This research program aims to understand the mechanisms and representations underlying the production and comprehension of referring expressions, with a focus on how comprehenders assess the accessibility of discourse entities and integrate this with lexical information as it unfolds over time. The experiments with adults examine a set of factors not previously identified with accessibility, including features of naturally-occurring language that are often not studied experimentally, like disfluency. These and other proposed experiments explore the role of expectancy in referent accessibility, as opposed to a traditional focus on how an entity has been treated in the history of the discourse. They take a multi-faceted approach, exploring how accessibility, expectancy and anaphoric form influence the processes of choosing and understanding referring expressions, and how this affects parsing decisions. The experiments with children ask basic questions about how they assess referent accessibility during on-line processing. Comprehension experiments with both adults and children will monitor participants' eye movements while they follow instructions to move objects on a display or listen to a narrative about a picture. Eye movements provide a fine-grained picture of the hypotheses that comprehenders make during on-line reference processing. The research with adults will also draw explicit links between language comprehension and production, by investigating the types of entities that get referred to frequently (and therefore may have a relatively high expectancy), and the referential forms that speakers choose under conditions of high and low expectancy. The proposed research will provide a systematic investigation of factors that affect referent accessibility, and how they affect reference comprehension and production. The study of reference comprehension and production is critical for understanding natural language use, which could not take place if comprehenders were not able to match the speaker's referring expressions with entities and concepts in the real world. The study of reference processing also bears directly on understanding other aspects of language processing, like syntactic ambiguity resolution and lexical access, which have recently been shown to be sensitive to the referential context. The development of explicit models of reference and other aspects of language processing is necessary for the diagnosis and treatment of language disorders, such as aphasia.
