Activation and use of script-based antecedents in anaphoric reference

This paper reports three experiments that investigated the accessibility of script-based antecedents during the resolution of anaphora. In the experiments, sentence pairs were used in which the referential tie between the sentences was either explicit (a repeated noun phrase), implied (the verb in the first sentence contained an empty case slot), or absent (the first sentence contained no referent). Additionally, the nouns of interest were either central or peripheral script-based concepts. Based on a model of the organization and activation of scripted knowledge developed in previous work ( F. R. Yekovich & C. H. Walker (1986) Journal of Memory and Language 25, 627–644), we argue that central concepts are part of the discourse model in working memory and thus are available as antecendents regardless of the text's features. In contrast, peripheral concepts depend on the text for their availability. Experiment 1 confirmed this by showing that comprehension time for central anaphoric sentences was independent of the type of referential tie used (i.e., Explicit = Implied = No Referent) whereas comprehension time for peripheral anaphoric sentences mirrored traditional text-based predictions (i.e., Explicit < Implied < No Referent). Experiment 2 demonstrated that this differential effect disappeared when the scripted context was not active. Experiment 3 extended the results of Experiment 1. When subjects were probed about the thematic relatedness of nouns immediately after reading the antecedent sentences, judgments about central concepts were fast and accurate regardless of the type of antecedent sentence. For peripheral concepts, judgment speed and accuracy depended on the type of sentence.