"Pragmatic Constraints on Narrative Processing: Actants and Anaphora Resolution,"

essay in progress by David Herman (North Carolina State University).

In this paper, I draw on linguistic pragmatics, discourse-model theory and structuralist narratology to outline an integrative approach to the problem of anaphora resolution in stories. Investigating the patterned distributions of anaphoric expressions in several narratives recorded during sociolinguistic interviews, my paper studies pragmatic (cognitively- and linguistically-based) constraints on narrative processing. The paper argues that anaphoric reference in narrative contexts needs to be examined as a special case of discourse anaphora; detecting and interpreting instances of narrative anaphora require models of human behavior that structuralist narratologists tried to characterize using the notion of narrative actants. Actants can be construed as labels for the most basic and general roles performed by participants in the narrated action. My analysis shows that the idea of actants, supplemented and enriched with later developments in linguistics and cognitive science, can throw light on the mechanisms of anaphoric reference in stories. At the same time, however, I argue that pragmatic and cognitive approaches to narrative discourse can gain in descriptive and explanatory adequacy if integrated with the earlier narratological research on actants.

The analysis is based on ghost stories told by inhabitants of several culturally and ethnically distinct areas of North Carolina during interviews designed to gather information about the dialects associated with these regions. (These interviews are part of a larger collection of tape-recorded interviews housed at the William C. Friday Linguistics Lab at North Carolina State University.) I begin from the premise that narrative is an interactional achievement; in particular, storytelling requires the collaborative construction of a discourse model by storyteller and story-recipient. Reliance on this (emergent) discourse model allows narrators to refer to items in the discourse via inexplicit and/or indefinite referring expressions, which are construed by story-recipients as informational updates on more or less enduring entities populating the storyworld. More generally, discourse models help storytellers and their interlocutors establish the existence of agents and objects in the storyworld, to track those agents and objects across contexts in which they lose or acquire properties, and to situate them in relationships with one another and with agents and objects belonging to other contexts. Thus, as a mental representation of the constituents of a given storyworld, a discourse model is what allows story-recipients to handle problems bound up with anaphora resolution. Yet narratively organized discourse presents processors with a unique type of anaphoric reference--and requires of them unique types of processing skills. In stories, it is not just that given information is called up by referring expressions signalling its status as given; more than this, story-recipients must be able to identify coreferential expressions in order to contruct an interpretation of where agents and objects are--and what they are doing--in the storyworld. Discourse anaphora in stories, then, needs to be studied as a communicative strategy used in the service of specifically narrative aims.

The present paper thus focuses attention on the way narrative, as a text-type, requires processing strategies different from those used to process other genres of discourse. After comparing the distribution of pro-forms and definite noun-phrases in narrative and non-narrative contexts, I show how storytellers prompt their listeners to construct discourse models in which expectations about human agency play an important role. Discourse models of this type are crucial for narrative comprehension; they support inferences about agents, events, and situations that could not otherwise be made. Actantial paradigms provide a useful vocabulary for describing these specialized discourse models. Furthermore, ghost stories as a narrative subgenre are especially interesting to study because they simultaneously depend on actantial paradigms and call them into question. Tales of the supernatural both evoke and undermine customary expectations about what human beings can and cannot do. Hence, studying anaphoric reference in ghost stories sheds light on the referential mechanisms--and the processing strategies--subtending narrative discourse at large. [D.H.]