Understanding anaphora: Rules used by readers in assigning pronominal referents∗

The research described herein was supported primarily by the Personnel and Training Research Programs, Psychological Sciences Division, Office of Naval Research, under Contract No. N00014–76‐C‐0461, Contract Authority Identification Number NR 154–386. and also by the National Institute of Education under Contract No. HEW‐NIE‐C‐400–76–0116. This project would not have been possible without the work of Marina Frederiksen, who wrote, to exacting specifications, the textual materials used in the experiment.

Abstract

The focus in this experiment was on the analysis of cohesive elements within a text and on the difficulty of their resolution within a particular text structure. The cohesive form we selected was a particular type of anaphoric reference—pronominal reference. The subjects' task was to read a text, sentence by sentence. The texts presented contained pronouns, and referents for the pronouns. In addition to reading the text, subjects were occasionally asked to report the correct referent for a pronoun that had appeared in the sentence they had just completed. With this probe task motivating them to analyze reference problems carefully as they were encountered, subjects' reading times were found to be closely related to structural properties of the text. Text variables of importance included the number of potential referents available, topicalization of the correct referent, staging of references to the correct or to alternative noun phrases, and the degree of ambiguity of the semantic constraints within the target sentence used in selecting the proper referent. The results support a reinstatement theory in which a number of available, potential referents are brought forward into working memory at the time a pronoun is encountered. The selection of a single referent from the set of potential referents is based upon a set of prioritizing rules that are sensitive to the staging of ideas within a text, and to features of surface syntactic structure, as well as to proposi‐tional content.

Notes

The research described herein was supported primarily by the Personnel and Training Research Programs, Psychological Sciences Division, Office of Naval Research, under Contract No. N00014–76‐C‐0461, Contract Authority Identification Number NR 154–386. and also by the National Institute of Education under Contract No. HEW‐NIE‐C‐400–76–0116. This project would not have been possible without the work of Marina Frederiksen, who wrote, to exacting specifications, the textual materials used in the experiment.