ABSTRACT
Three self-paced reading experiments investigated the nature of the representation in which pronouns and other NP anaphors find antecedents. The first experiment provided no evidence that the representation consisted of syntactic constituents, and was consistent with the widely-held opinion that antecedents for pronouns are found in a representation that captures referential identity. The second and third experiments examined the effects of topichood and distance between antecedent and pronoun on pronoun reading time. They demonstrated that pronouns with topic antecedents were read quickly, regardless of distance. However, closer analysis of the data indicated that some sentences seemed to have more than one equally quickly accessed antecedent, in line with suggestions that discourse centres, not topics, are preferred antecedents for pronouns.
