Dissertations, Theses, and Capstone Projects

Date of Degree

1989

Document Type

Dissertation

Degree Name

Ph.D.

Program

Psychology

Advisor

Martin S. Chodorow

Committee Members

Stanley Novak

Virginia Valian

Subject Categories

Psychology

Abstract

Through a set of two experiments, the present study attempted to determine which of three language processing strategies, that is, serial, parallel or delay is employed in parsing two kinds of structurally ambiguous visually presented sentences (transitive and verb complement). The study used a relatively new technique, a self paced syntactic decision task whose sensitivity to local parsing complexity was demonstrated in the first experiment through a partial replication of Ford's (1983) work with relative clause sentences. The findings showed Object relatives to be harder to process at the position of the main verb. The same findings from a followup experiment in which the relative clause was lengthened lent support to a Filler-gap explanation of parsing Object relatives proposed by Ford (1983). Also found in the first experiment was weak support for either a parallel or delay parsing of ambiguous transitive sentences and strong support for a parallel parsing of verb complement sentences. As followup, in the second experiment, an attempt was made to differentiate the parallel from delay parse of transitive sentences. In addition, an examination of the parsing strategy used for verb complement sentences with lengthened ambiguous regions was motivated by Kurtzman's (1984) work. Results of Experiment 2 suggested a mixed parsing strategy for the processing of verb complements with lengthened ambiguous regions. The findings for modified transitives suggested a serial parsing strategy which was unexpected given findings from the first experiment and findings by Kurtzman (1984). Biasing by lexical preference was deemed unlikely but not definitively ruled out. The strength of lexical preference (i.e., verbs biasing parsing strategies) was not found to be significantly related to the strength of garden pathing (i.e., degree of difficulty reassigning NP's from a direct object to complement subject structure) in complement clause sentences. A second methodological variable (i.e., grammaticality judgement errors) was examined and ruled out as an alternative measure of local parsing complexity. Post hoc analysis of standard errors from ambiguous and unambiguous sentences raised questions regarding differentiation of parallel processing from mixed subject strategies, which should be addressed in future work.

Comments

Digital reproduction from the UMI microform.

Included in

Psychology Commons

Share

COinS