Small Worlds of Discourse and the Spectrum of Accommodation: A Psycholinguistic Investigation of the Accommodation Entailed by the Definite Article when its Determinant is Discourse-New

I conducted three experiments in which I investigated how people use the definite article with entities that have never been explicitly introduced. I claim that using the definite article to determine a particular entity presupposes that that entity has a preestablished discourse referent. If an explicit discourse referent has not been supplied by previous context, additional presuppositions must be accommodated to infer one instead. In experiment 1, participants were given a self-paced reading task to determine if less felicitous accommodations took longer to process than more felicitous accommodations. I compared both supported and un-supported inferences of a target discourse-new entity to contexts in which the entity had already been explicitly mentioned. I found that prior mention of the target entity made subsequent processing times of the same entity significantly faster, and well-supported accommodations were marginally faster to process than poorly-supported accommodations. In experiment 2, I investigated the requirements of the definite article by asking participants to rate the felicity of two-sentence pairs in which one entity was singled out with the definite article, from a set of equally plausible entities. In one condition, context made one entity clearly more salient than others of its type. In the second condition, all entities were equally salient. I found that when the target entity was singled out over others of its type, items were rated significantly higher than when there was no reason to single out one entity. This indicates that in order to use the definite article, context must supply the designated entity with a single, unambiguous discourse referent with which it can link. This means that the entity must be both relevant and more salient than all others of its type. I call this view the Small World hypothesis. It resolves the traditional distinction between uniqueness and identifiability with the concept of discourse referents, because in order for the entity to have a single unambiguous discourse referent candidate, the target entity must both be identifiable and locally unique. In experiment 3, I conducted a forced-production study. I had participants supply either the definite or indefinite article to contexts that met the requirements of the Small World hypothesis by varying degrees. I found that in contexts that met the requirements of the Small World, participants consistently chose the definite article. In contexts where the requirements were only partially met, participants chose either the definite or the indefinite. In contexts where the requirements were not met, participants consistently preferred the indefinite article. I combined the findings of experiments 1, 2 and 3 to motivate the Spectrum of Accommodation, in which I claim that when using the definite article, the presuppositions necessary to treat discourse-new entities as discourse-old ones are accommodated with relative ease or difficulty, depending upon how well context manages to make a particular entity relevant and more salient than all others of its type, by linking it up to an unambiguous discourse referent.