Children's learning of language-universal and language-specific principles of argument representation was the topic under investigation in the three studies comprising this thesis. Another objective was to investigate whether a discourse-pragmatic approach could be employed to explain children's patterns of argument omission and production, developmentally and crosslinguistically. To answer these questions, referential choice in the spontaneous language of monolingual English-speaking and monolingual Japanese-speaking children and their mothers was developmentally investigated whereby a sentence argument's morphological form (null, pronominal, lexical), referential status (given, new), and syntactic location (transitive subject, transitive object, intransitive subject) were systematically analysed. The first and second studies revealed that neither the English-speaking nor the Japanese-speaking children showed sensitivity to the referential distinction between given and new information early on in development (at 21 months of age). The English-speaking children mastered English-specific referential conventions between MLU 2.00 and 3.99 (between 24 and 32 months) and employed non-linguistic pragmatic correlates to supplement unconventional argument use from as early as MLU 1.00 (between 21 and 23 months). By contrast, the Japanese-speaking children showed unconventional referential choices as late as MLU 4.00 (between 33 and 36 months), as well as inconsistent use of non-linguistic pragmatic correlates. The third study revealed that, although language-specific differences were observed, neither group of children violated any of the four Preferred Argument Structure (PAS) constraints: The children avoided using more than one new or lexical argument per transitive clause and avoided casting new or lexical arguments as transitive subjects. However, evidence of sensitivity to PAS strategies from early on in development was inconclusive because the children omitted most sentence arguments at the beginning of speech production. Finally, all three studies revealed that children's referential choices that were inconsistent with expected discourse-pragmatic principles reflected similar patterns observed in parental input. Altogether, this set of studies led to the following general conclusions regarding the learning of argument representation and distribution in syntax: (1) a discourse-pragmatic approach can explain language-universal features of argument omission and production in child language and (2) language-specific strategies are learned via parental input.

No description available.

Thesis (Ph. D.)--University of Hawaii at Manoa, 1995. / Includes bibliographical references (leaves 273-280). / Microfiche. / xiii, 280 leaves, bound 29 cm