The characterization of the relationship between morphological and syntactic properties of sentences is central to current linguistic theories, for example, suggesting that morphological features play a central role in determining verb movement and the occurrence of null and expletive subjects. Given such a theory of competence, many questions arise about the role of these features in performance, such as: Is there a primary role for morpho-syntactic feature information in parsing sentences? How is feature checking accomplished and at what stage of the parse? The thesis addresses these questions with on and off-line studies of the comprehension of subject-verb agreement in English. Two general views are contrasted: (1) agreement information constrains initial parsing decisions (the “Structure Building” hypothesis) and (2) agreement is checked on structure which has been built using only major category information and structural parsing principles (the “Structure Evaluation” hypothesis). Evidence presented in this thesis, as well as previously reported results, are argued to support the Structure Evaluation hypothesis. A detailed statement of the checking process (the Structural Agreement Check) is developed within the Structure Evaluation approach. The SAC assumes (following previous work) that the nominal singular is underspecified and that feature checking is initiated at the verb. Evidence is presented that checking is enforced by passing agreement features through the links of the phrase structure tree between the subject and verb. Further evidence in favor of the Structure Evaluation hypothesis and the SAC will come from on- and off-line data in three previously uninvestigated domains: agreement with ambiguous Pseudo-partitive DPs, agreement with Pseudo-partitive and Coordinate DPs in verb-subject order constructions and agreement in more complex structural contexts (subject-verb agreement embedded in a filler-gap feature dependency and verb agreement in ambiguous subject relative clauses). In addition to providing a more detailed characterization of the role of agreement features and representation of the agreement relation in parsing, the results contribute to the analysis of the structure of Pseudo-partitive DPs and the Existential construction (as compared to other post-verbal subject constructions). Finally, it is shown in two case studies that the processing mechanisms proposed here may play a role in destabilizing perceivers' intuitions about grammaticality and possibly lead to language change.

This study is about the interpretation of a plural pronoun in English. A two-stage performance model is proposed on the basis of Link's competence theory which claims that plurals denote sums of individuals, i.e. groups. The first stage is the construction of group discourse entities on syntactic and non-syntactic grounds. Construction on syntactic grounds is described by the NP rule which states that an N P with a plural feature triggers a group discourse entity. Construction on non-syntactic grounds is described by the equivalence hypothesis which states that if two individuals are equivalent with respect to some property, they can be collected as a group. The second stage concerns the selection of the preferred antecedent. I proposed the closure strategy which states that if all the individuals in the interpretation domain belong to one group, select this group as the antecedent of a plural pronoun. One crucial prediction of this model is the preference for the maximal group in the interpretation domain. Experiments 1, 2 and 3 tested whether symmetric predicates lead to the construction of a group. Experiments 4 and 5 tested whether ontological knowledge influences the construction of a group. Experiment 6 tested whether perspective influences the interpretation of a plural pronoun.

Source: Masters Abstracts International, Volume: 46-06, page: 3258. Adviser: Scott T. Walters. Includes bibliographical references.