Computational Pragmatics

Instructor(s): Christopher Potts

We will identify and explore ways in which linguistic pragmatics can be informed and enriched by computational methods, and ways in which work in applied and theoretical natural language understanding can benefit from the insights of linguistic theory. The topics will include conversational implicature, especially in connection with enriching answers in context, presupposition, discourse anaphora, and sentiment analysis. The lectures will be highly interactive, with a daily mix of lecturing and hands-on problem solving. (Laptops are highly encouraged.) The course is introductory, but some background in pragmatics and computation is required; see the prerequisites for details.

A course in semantics, pragmatics, or natural language understanding, or simultaneous enrollment in one of the Institute courses in semantics and pragmatics; and enough programming experience (in any language) to, say, iterate through 100,000 text files extracting counts of things from their contents and formatting the results.