Keynote Speech (Ⅱ)

摘要

A short, ambiguous message can convey a lot of information, provided the listener is willing to make inferences based on assumptions about the speaker and the context of the message. These sorts of pragmatic inferences are critical in facilitating efficient human communication, and have been characterized informally using tools like Grice's conversational maxims. In this talk, I'll describe our work on a new, probabilistic framework for referential communication in context. This framework shows good fit to adults' and children's judgments across many experiments, provides extensions to a variety of complex linguistic phenomena, and resolves some important puzzles about language processing. I'll end by describing how we have begun to test this framework using data from large-scale corpora of social media conversations.