9319780 MYERS-SCOTTON ABSTRACT This a project to investigate the structural constraints on intrasentential codeswitching. Specifically, the investigators will collect and analyze intrasentential codeswitching in naturally-occurring languages in particular language pairs, some in the United States and others in Africa or Europe. The language pairs have been selected because apparent differences in the structure of their lexical items will test the hypotheses contained herein. While this study will further our knowledge about intrasentential codeswitching, it is more important for the information it will provide about the nature of lexical entries in some universal sense. This information becomes available from codeswitching data because the existence of sentences containing codeswitching seems to depend on compatibilities between the languages involved regarding Lexical-Conceptual Structures and Predicate-Argument Structures. The playing out of these compatibilities in codeswitching, in turn, provides a uniquely available 'empirical window' on the viability of key theoretical claims about the structure of language. The proposal's eight hypotheses extend the analysis found in the Matrix Language Frame Model of codeswitching (Myers-Scotton 1993b). Under this model, one of the languages participating in codeswitching is more dominant (the Matrix Language) in structuring codeswitching utterances. The less dominant language is called the Embedded Language. The distinction between content and functional categories is also critical in this model. While that model successfully predicts the form of Embedded Language material in sentential frames prepared by the Matrix Language, it is not sufficiently constrained to explain why one option rather than another occurs and the form which Embedded Language material takes. The hypotheses in this proposal provide necessary constraints. This research will provide evidence for or agai nst theories about the nature of the abstract features of lexical entries, including semantic content, and their relation to actually occurring forms. In addition, the results may provide motivations for language production modeling in psycholinguistics, especially in regard to the nature of lemma entries in the mental lexicon. Finally, the project has a material objective. It will provide a large-scale computerized data base of diverse codeswitching data. It is expected that collecting preliminary analysis, data base entry, formal analysis, and a presentation of the results of this study will require three years. ***
