On-line Reference Assignment for Anaphoric and Non-Anaphoric Nouns: A Unified, Memory-Based Model in ACT-R - eScholarship

On-line Reference Assignment for Anaphoric and Non-Anaphoric Nouns: A Unified, Memory-Based Model in ACT-R Aryn Pyke (aryn.pyke@gmail.com) Institute of Cognitive Science, Carleton University, 1125 Colonel By Drive Ottawa, ON K1S 5B6 Canada Robert L. West (rlwest@connect.carleton.ca) Jo-Anne LeFevre (jlefevre@connect.carleton.ca) Institute of Cognitive Science, 1125 Colonel By Drive Ottawa, ON K1S 5B6 Canada Department of Psychology, 1125 Colonel By Drive Ottawa, ON K1S 5B6 Canada Sengul, 1979; Kintsch & van Dijk, 1978; O'Brien, Plewes, & Albrecht, 1990). In such strategic-search models, the discourse might be mentally represented as a proposition network, and the reader might systematically troll backwards through it in search of the antecedent that (according to some criterion) could constitute a match to the current anaphor term. After all, how else could we account for the fact that readers come across the right referent (i.e., the particular one mentioned earlier in the discourse)? An answer to this “How else” question is furnished by the memory-based view of text processing (see Gerrig & O’Brien, 2005 for a review). According to the memory- based view, the successful retrieval of a referent need not require (or constitute evidence of) a strategic, proactive search, because passive general-purpose memory processes often can automatically bring the intended referent to mind. In particular, under the resonance model (e.g., Gernsbacher, 1989; Myers & O’Brien, 1998) current information in working memory (i.e., the anaphoric noun) serves as a cue that automatically boosts activation of other entities throughout long-term memory -- including, ideally, the intended referent -- in accord with their conceptual overlap with the cue. Thus, at the time the anaphor ‘fruit’ in (2) is encountered, the apple referent can be automatically re- activated via resonance in virtue of its conceptual overlap with the anaphor (a pre-existing conceptual association). Certainly higher-level and pragmatic processes may also play a role in comprehension. However, to account for readers’ frequent success at referent reactivation during (first-pass) anaphor processing, we agree that there may be “no need to invoke any process other than general memory processing” (Gerrig & O’Brien, 2005, p. 230). The computational model in the present paper constitutes an existence proof that memory-based models are indeed sufficient, not only in principle, but in practice, to account for a high rate of success at first-pass referent retrieval. The present paper and model also identify and address a fundamental, but we believe, previously neglected and under-estimated problem: The Anaphoric Classification Problem. In particular, how (and how accurately) can readers judge whether or not a noun is anaphoric during first-pass processing? Our model demonstrates how the memory-based view can be operationalized to address this classification problem, and in particular, to predict when a Abstract The computational model in present paper confirms that memory-based accounts are sufficient to account for a high rate of success at first-pass referent retrieval for anaphoric (and non-anaphoric) nouns. Because even definite noun phrases can often be non-anaphoric (e.g., Poesio & Vieira, 1998), an adequate model must account for how a reader makes an explicit or implicit decision about the anaphoric status of a noun (herein: The Anaphoric Classification Problem). We explain why we are inclined to reject the conventional intuition that: the failure to find/retrieve a referent within the discourse then, serially, leads to treating a (possibly anaphoric) noun as a new referent. Instead, we extend the memory-based account to address this classification problem. We suggest that LTM contains both generic referent types and specific referent tokens, which simultaneously compete for retrieval via resonance. The nature of what is retrieved (token vs. type) determines whether the reader effectively treats a noun as anaphoric or not. Our model predicts whether an anaphor in a given text will be misinterpreted as a new referent during first-pass processing. The influence of anaphor word choice is explained, and encompasses metaphoric anaphors. Keywords: noun anaphora; memory-based text processing; resonance; reference assignment; cognitive modeling; ACT-R Introduction An anaphoric noun is one that denotes a referent that was previously mentioned in the discourse, but possibly using a different antecedent term. For example, in (2) “fruit” (or “apple”) is used anaphorically to denote the referent introduced by the antecedent “apple” in (1). (1) John bought an apple. (2) John enjoyed the fruit/apple. Readers are often able to re-activate the intended referent 1 almost immediately after encountering an anaphoric noun (Dell, McKoon, & Ratcliff, 1983; Sanford & Garrod, 1989) that is, after the first-pass processing of the noun. To account for this on-line ability, some models suggest that when a reader encounters an anaphoric noun, he/she undertakes a strategic search for an antecedent through a representation of the discourse context (e.g., Clark & The term ‘referent’ is being used here in the cognitive sense (as in Gundel, Hedberg, & Zacharski, 2001) to mean the mental representation of the entity (person or object) in question.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.