Type Raised Children: Extending Categorial Grammar as a Theory of Acquisition
AuthorDrozd, Kenneth F.
AffiliationUniversity of Arizona
MetadataShow full item record
DescriptionPublished as Coyote Papers: Working Papers in Linguistics from A-Z, Unification Based Approaches to Natural Languages
AbstractAcquisitionists in the 1960's and 1970's have interpreted the semantic and syntactic patterns evident in children's two -word combinations under one of three analyses. These consist of (i) combinations of closed classes of constant terms (pivots) with a set of open class forms (Braine 1963), (ii) a set of surface structures reduced by transformations from underlying structures displaying semantic relations (Bloom 1970), or (iii) a child's attempts to find positional patterns for salient conceptual categories (Bowerman 1973; Braine 1976). Each of these approaches sets out to explain an isolated set of semantic or syntactic regularities observed in childrens' early two word utterances. Missing from each of these analyses, with the exception of Bloom (1970), is a description of how the syntax and semantics of these utterances are related. In addition, all of these analyses neither describe what kind of linguistic knowledge children use in creating two -word utterances or how this knowledge prepares children to be able to create more complex utterances. Consequently, each analysis fails to characterize the production of two-word utterances as a necessary stage in the development of adult linguistic competence. A categorial approach to acquisition offers a precise representation of the linguistic knowledge underlying these early lexical combinations as well as of the stages a child experiences during the language acquisition process. Categorial grammars assume that combinatorial possibilities for linguistic categories, restricted to being either functors or arguments, are specified in their lexical descriptions rather than in phrase markers or subcategorization frames familiar from phrase structure grammars. Linguistic competence is defined, in part, as knowledge of linguistic functions and the range of arguments they may apply to. Lexical categories are partially ordered by their complexity, a notion which provides a procedure for determining the sequence of stages of language acquisition. The linguistic knowledge responsible for creating early two-word combinations, then, is defined as knowledge of an elementary order of categorial complexity which is also required for creating and interpreting more complex utterances of the language. Lexical categories and their combinations are assigned corresponding semantic values, thereby offering a principled method of simultaneously encoding both the semantic and lexical properties of categories used by children, a method which is unavailable in current phrase structure accounts. In sum, the categorial analysis proposed in this paper makes interesting and specific predictions about the nature of children's grammatical competence as well as the nature of the acquisition sequence. In this paper, I am concerned with showing how a categorial grammar can be manipulated to describe the syntactic patterns underlying Pivot Grammars (Braine 1963) as well as those patterns implicit in children's first attempts at combining nouns, verbs, adjectives, auxiliaries and additive conjunctions cross-linguistically. In Section 1, I present Braine's (1963) original description of Pivot Grammars as well as evidence against Braine's analysis made by Bloom (1970) and Bowerman (1973). 1 propose a categorial grammar which describes the properties of pivot constructions cross-linguistically and which places pivot constructions within the acquisition sequence as a distinct utterance class. The acquisition sequence is based on the notion of categorial complexity, a measure by which children may infer complex categories from simpler ones already productive in the grammar. In Section 2, I extend the categorial analysis to include those utterance types incorrectly predicted by Braine's analysis not to occur at the pivot grammar stage, namely V+N combinations and N+N combinations encoding possession and location (Bloom 1970, Bowerman 1973). 1 show that this class of utterances, which has different properties than pivot constructions, is nonetheless a product of the same order of complexity attributed to Pivot Grammars. ln Section 3, 1 show that the sequence of auxiliary and additive conjunction acquisition corresponds to the sequence of orders of complexity predicted by the categorical grammar of the first section. I summarize the results of my analysis in the final section.