245 Actes de l'ACL 2002/ 2002 CLA Proceedings SYNTACTIC FEATURES AND DISCOURSE FACTORS IN CHILDREN'S INTERPRETATION OF DEFINITE DETERMINERS IN INALIENABLE POSSESSIONS * Ana T. Pérez-Leroux University of Toronto Cristina Schmitt Alan Munn Michigan State University 1. Introduction One prevailing assumption within generative approaches to language acquisition is the continuity hypothesis, which states that learners’ intermediate stages are possible adult grammars. As the theoretical construct of parameter has evolved towards a smaller, narrowly lexical view of parametrization, the field’s perspective on the task of learning is changing as well. One important domain of language variation (and thus a central part of the learning task) pertains to how languages partition semantic spaces— how the relevant semantic features are mapped onto the relevant morphosyntactic categories. We assume that there is a universal vocabulary of interpretable features available to children’s language acquisition device. A second assumption about learning properties of the syntax/semantic interface is that forms with comparable morphosyntactic distribution lexically compete for a given semantic space (forms avoid being given identical senses, as in Pinker 1984). These assumptions define the learning task as a bootstrapping process that narrows the set of syntactically relevant interpretable features to those of the target grammar. Form-sense mappings are established when lexical units are associated to specific syntactic and semantic frames. Within the domain of definite determiners, crosslinguistic differences in the syntactic and semantic distribution of determiners constitute part of the learning problem and may give rise to non-target representations in young children. Children learning languages with comparable definite determiners may initially develop the same representation for these items, but can eventually use semantic and syntactic distributional information of competing items to narrow the domain * This research was conducted with partial support from the Faculty of Arts and Sciences at the University of Toronto. We also thank Michelle Deirish for her kind assistance with child data collection, Jennifer Cabrelli, Boonjeera Chiravate, Carolina Holtheuer, Sarah Kurzhals and Karen Miller for adult data collection and Tom Roeper, Jill deVilliers, Sergio Baauw and Eliane Ramos for helpful discussion and suggestions.