Abstract Holler & Irmen
Is Anaphor Resolution Constrained by the Right Frontier?
There is ample evidence that the accessibility of information is influenced by various lexical and syntactic factors (cf. Klin & Guzmán & Weingartner & Ralano 2006). The influence of discourse-structure on the salience of information, however, has not been in focus of psycholinguistics, yet. This differs from the situation in theoretical linguistics where recent research is directed at the discourse-structural conditions constraining the salience of text information. One of the most influential frameworks is Segmented Discourse Representation Theory (SDRT, Asher & Lascarides 2003). Its fundamental assumption is that discourse consists of a set of discourse units, which are connected by certain discourse relations. SDRT offers a formal account of the hypothesis that discourse has a hierarchical structure upon which interpretation depends. One of the central constraints on discourse update and anaphor resolution in SDRT is the Right Frontier Constraint (RFC), originally proposed by Polanyi (1988). It says that an antecedent for an anaphoric expression is accessible only at the right hand side of any level of a linearly ordered discourse parse tree. The study presented here investigates inasmuch discourse structural information affects the way intersentential anaphora are resolved. It aims at an empirical verification of the RFC. Following this constraint, it is hypothesized that potential antecedents of an anaphor that are placed at the right frontier of a discourse unit can be accessed more easily than antecedents placed somewhere else. We examined this hypothesis in a questionnaire study.
Experimental passages consisted of six lines that contained a pronominal anaphor in the last line and two potential antecedents in the preceding text, one introduced in the first, one in the fourth line. The relative position of antecedents to the right frontier was manipulated through the discourse relation between first and second antecedent as well as through the filler information interposed between second antecedent and the anaphor. Type A: The first antecedent stood at the right frontier; Type B: Only the second antecedent stood at the right frontier; Type C: Both potential antecedents stood at the right frontier. Within each type the two potential antecedents either had the same or different grammatical gender. Participants were asked to name the antecedent of the pronominal anaphor by answering a question that rephrased the last line of a passage.
Following the RFC participants should tend to chose the first antecedent in Type A and the second in Type B. No such difference is expected if other factors are more influential. Recency should, for example, always favor the second antecedent, gender congruency the first antecedent in case of unequal gender antecedents.
Frequencies of choices were compared between the three different types of experimental passages. In case of unequal genders participants almost always chose the first antecedent, which was gender congruent to the anaphor. Thus, no effect of discourse relation could be found here. In case of equal gender antecedents, participants’ choices differed between types of items and were in line with the idea of a right frontier constraint. The present study represents a fruitful attempt to derive testable predictions from a linguistic theory on discourse relations. Results indicate that the RFC does not hold unrestrictively. Morpho-syntactic information, for example, may override the RFC.
Asher, N. and A. Lascarides (2003). Logics of Discourse. Cambridge UP.
Klin, C. M., Guzmán, A. E., Weingartner, K. M. and A. S. Ralano (2006). When anaphor resolution fails: Partial encoding of anaphoric inferences. Journal of Memory and Language, 54, 131-143.
Polanyi, L. (1988). A formal model of the structure of discourse. Journal of Pragmatics, 12.