Janet Fodor will present “What parsers want from grammars” in the seminar hub at 3:30 on Friday, November 14. An abstract follows. Professor Fodor will be at UMass all of Friday. If you’d like to schedule an appointment to see her, get in touch with Stefan Keine.
From the earliest days of transformational generative grammar, there has been an uncomfortable truce between the formal study of syntactic structure and the development of models of human sentence processing. Processing models are unable to make practical use of formal derivational operations in assigning structure to incoming word strings. With the advent of Chomsky’s Minimalist Program (MP) the situation has worsened. Syntactic derivations have been revised, on theoretical grounds, so that both structure building and movement are now misaligned with parsing. MP derivations inherently operate bottom-up, which for right-branching constructions means from right to left. Taken literally, this would imply that parsing begins at the end of a sentence. After noting a flurry of reactions to this impractical conclusion (rejection of the problem by Neeleman & van de Koot; proposed solutions to the problem by Fong, Chesi, and den Dikken), I will take the viewpoint of a working psycholinguist and propose instead that an efficient parser builds MP trees left-to-right and top-to-bottom, from interlocking chunks of tree structure. Where do the chunks come from? The MP grammar generates complete sentential trees (bottom-up, right-to-left – no problem!) which are then chopped into the parser-friendly building blocks.