Abstract: Linguists and psycholinguists interested in discourse often appeal to the concept of information structure, and particularly to the distinction between a sentence’s given and new information. In our recent work, we have argued that these two information types are processed differently during comprehension. Given information undergoes shallow processing: all the comprehension system needs to do with the given portion of a sentence is verify that the concepts seem familiar and that they can be related to previous discourse. This shallow processing strategy for old information allows the system to save its processing resources for the integration of new information, which undergoes deeper syntactic and semantic processing. This deeper processing is used to integrate the new content into the ongoing discourse representation and to anticipate upcoming forms. Our proposal, then, is that good enough processing is reserved for the given or presupposed part of a sentence, and that prediction is a mechanism to facilitate the integration of new or focused information. This approach brings together three different literatures that have tended to be treated independently: discourse structure, prediction, and good enough processing. In this presentation, I will present evidence for this approach and attempt to relate the proposals to broader issues concerning discourse and text structuring.
Abstract: It has become fairly widely accepted that pragmatic inferencing can be a factor in language change. During the 1990s I developed hypotheses about the role of invited inferences in change, especially grammaticalization. These neoGricean hypotheses were brought together as the Invited Inferencing Theory of Change (IITSC) (Traugott and Dasher 2002). Challenges to the IITSC were advanced (e.g. Hansen and Waltereit 2006) and to the neoGricean perspective, especially the notion of conversational implicature in general (e.g. Lepore and Stone 2015). Nevertheless, the neoGricean perspective persists, modified by recent advances in pragmatics and semantics (Horn 2016). I reconsider the IITSC in the light of constructionalization and of discourse structuring and turn-taking as evidenced in interactional historical texts. I suggest that at least three types of inferences play a role in interactional contexts: local inferences associated with specific expressions; discourse structuring inferences pertaining to factors like coherence, backgrounding and foregrounding; and turn taking inferences associated with turn relevant positions. These types of inferences are illustrated with an analysis of the development of the digression marker by the way, with focus on its use in topic-shifting.
Haut de page