Workshop on Natural Logic, Proof Theory, and Computational Semantics |

Logic was originally meant to formalize and analyze arguments in
natural language. But in the 20th century the main developments in
logic focused on mathematics and its foundations. As far as natural
language goes, logic has been, as Mark Steedman puts it in the preface
to his recent book, like a tailor who when asked by a customer
(Linguistics) to provide trousers, says it can only produce jackets,
which are clearly topologically equivalent. Perhaps, but not as useful,
especially if you want to move about.
Currently, the need for natural language understanding by computers provides an impetus for the field to refocus on systems that formalize the inferences that can be drawn on the basis of natural language statements. This has led systems tuned to natural language semantics to reconnect with the older tradition. The logical and conceptual underpinnings of some of these systems remains unclear, although some recent work has begun to address formal foundations. The aim of this workshop is to contribute to this direction in semantics and to discuss logics, especially proof systems, well-suited for natural language semantics and to explore comparisons between these systems, including: (a) the study of fragments of first-order logic which are powerful enough to represent interesting linguistic phenomena and yet small enough to be computationally feasible; (b) logical systems for reasoning about polarities; (c) extended syllogistic logics; (d) logics and algorithms for use in textual entailment; (e) proof theoretic semantics; (f) formalizations of inference tasks centered on linguistic expressions; (g) theorem proving for textual entailment tasks. |