This paper examines the role that linguistic and cognitive prominence play in the resolution of anaphor–antecedent relationships. In two experiments, we found that pronouns are immediately sensitive to the cognitive prominence of potential antecedents when other antecedent selection cues are uninformative. In experiment 1, results suggest that despite their theoretical dissimilarities, topic and contrastive focus both serve to enhance cognitive prominence. Results from experiment 2 suggest that the contrastive prosody appropriate for focus constructions may also play an important role in (...) enhancing cognitive prominence. Thus different types of linguistic prominence (topic, contrastive focus) appear to have the common effect of increasing the cognitive prominence of the discourse referent. For pronouns with two possible antecedents, the cognitive prominence of an antecedent aids in anaphor resolution, immediately biasing selection towards the more prominent (and ultimately preferred) antecedent. (shrink)

One area of language research that has received a great deal of attention, both theoretical and empirical, is the use of anaphoric expressions. Such expressions can be thought of as serving two functions: the primary function is to refer back to a referent from previous discourse, and the secondary, but no less important, function is to help provide discourse coherence and structure. Third person pronouns such as he or she are anaphoric expressions par excellence, but fuller anaphoric expressions, including demonstrative (...) and definite noun phrases (NPs) such as that woman and the woman are also used in natural discourse. In this chapter we shall focus primarily on issues concerning definite NP anaphor resolution, and in particular we shall examine the interaction of two factors that are related to the identification of antecedents: the focus status of the antecedent and the semantic relationship between the antecedent and the anaphor (including semantic overlap). After presenting these factors, we will discuss one particular approach to anaphor resolution, Almor's (1999) Informational Load Hypothesis (ILH), and present three experiments that examined the findings presented in Almor (1999). The results of these experiments will lead us to consider in more detail the secondary, discourse-structuring function of anaphoric expressions. (shrink)

This chapter presents a new model of coreferential NP anaphora processing, JANUS, within the mental models framework. It summarises previous research on NP anaphora that is most pertinent to JANUS, and outlines two previous attempts to provide an integrated theory of NP anaphora: Centering Theory and Almor’s Informational Load Hypothesis. Each has it problems, but the Informational Load Hypothesis is more firmly rooted in psychology, and closer to our own approach. JANUS incorporates many ideas from the Informational Load Hypothesis, but (...) attempts to address its problems. JANUS assumes that the semantic content of an anaphor should be justified by two types of role that the anaphoric expression plays. Backward-looking roles are primarily concerned with identifying the antecedent (and referent) of the anaphor. Forward-looking roles relate to what is to be said about the referent in upcoming discourse. These two types of roles give JANUS its name. (shrink)