An investigation of theories of focusing

Abstract

This thesis investigates the proposals of a number of psychological and computational accounts of focusing and pronoun interpretation in view of obtaining experimental evidence for the questions raised by the accounts. Four distinct but related studies were conducted with the aim of bringing together various frameworks as a step towards developing an integrated model of the processes of comprehension. Experiments 1-3 in Chapter 2 show that thematic role and surface position focusing take precedence over the salience from naming, but that naming effects are seen in the absence of thematic role focusing. Experiments 4-5 in Chapter 3 show the effect of clause subordination in certain complex sentences, with main clauses being more prominent than subordinate ones. Experiments 6-8 in Chapter 3 show that this effect may not be generalised to different types of complex sentences, however. Experiments 9-18 in Chapter 4 show that animacy has a strong effect on prominence, overriding thematic role and surface position effects. The presence of these latter two effects is crucially dependent on the pattern of animacy. Experiments 19-21 in Chapter 5 show the effects of grammatical parallelism, in which the features of both the anaphor and the antecedent have an influence, which overlays structural focusing. These results show that a variety of constraints can complete in determining the accessibility of discourse referents. The structural, semantic, and pragmatic discourse context in which referents are introduced and the attributes of the cue used to re-access them have a role. The findings are discussed in terms of an activation-based framework, whereby pronoun resolution is determined by the relative activation of the potential antecedents in the mental representation of the discourse. They suggest a dynamic model of focusing in which an antecedent’s features establish and update the focus, and in which certain linguistic elements may trigger the current focus to be modified.