talk by Noah Goodman about “Uncertainty in language and thought”

 "I’ll describe how probabilistic programs — a formal merger of programming languages and statistics — let’s us make powerful models of commonsense human knowledge and reasoning. I’ll then focus on using these tools to describe natural language usage, particularly the pragmatics of communication. I’ll talk about scalar implicature (“some” implies "not all”), vagueness for adjectives (“bob is tall” is ambiguous and context-sensitive), and hyperbole (“he talked for ten thousand minutes” conveys less time, but a strong opinion!)."