The Effect of Subordination on Antecedent Prominence in Anaphora Processing

Abstract

One important factor affecting anaphor resolution is the prominence of its antecedent in the discourse. Discourse Prominence Theory (Gordon & Hendrick, 1998) defines antecedent prominence in terms of its height in the phrase structure tree. Two reading time experiments were conducted to test this theory by investigating the effect of subordination on antecedent prominence both within and between sentences. Both experiments had a 2×2 design with Structure (intra- vs. inter-sentential) and Anaphor (pronoun vs. repeated name) as independent variables, and differed in the degree of subordination of the antecedent. The results of the experiments show a main effect of Anaphor such that clauses/sentences containing pronouns were read faster than those containing repeated names, but no interaction between Structure and Anaphor. These results indicate that the repeated name penalty occurs no matter how deeply the antecedent is embedded within subordinate clauses. Taken together, the findings demonstrate that antecedent prominence is not determined by its height in the phrase structure as claimed by Discourse Prominence Theory but rather may be related to the syntactic function (subject) and status (head noun) of antecedents.