Chicago Press) Reviewed by

Literature and Cognition applies methods and ideas from computational linguistics and cognitive science to literature. It explores a wide range of topics, and is intended to be read by literary theorists as well as by computational linguists and cognitive scientists. From the perspective of computational linguistics, the largest contributions are to present Hobbs's previously published theory of discourse coherence in the context of an overall framework for discourse interpretation, and then use it to analyze a sonnet in great detail. (Hobbs also analyzes a novella in detail, but much less directly in terms of the theory.) These will be the focus of this review. The chapter presenting the overall framework (Chapter 3) is entitled "A theory of discourse interpretation, " but it outlines what would constitute such a theory, rather than presenting an actual theory itself. A major theme of the presentation is the au-thor's view that a theory of discourse interpretation "must first and foremost be a theory of how knowledge is used in solving the interpretation problems posed by the discourse " (p. 41). Such a theory would include six subtheories. The first four are famil-iar: the knowledge-representation language; the encoding of background knowledg