Nonmonotonicity in Linguistics

Publisher Summary This chapter outlines the logical issues of nonmonotonicity, with pointers to the literature, to indicate some of its applications in linguistics and provides references to the work that has already been done in applying nonmonotonic logic in linguistics. Nonmonotonicity is a property of the consequence relation, and a system of logic is said to be nonmonotonic if its consequence relation has the nonmonotonicity property. Nonmonotonicity is intimately connected with default inferences. The need to formulate and investigate nonmonotonic logics arose out of a recognition that other forms of reasoning are not always subjected to this constraint, together with a reluctance to explain away the nonmonotonic effects by using mechanisms, such as probability, that are orthogonal to logical consequence. Nonmonotonic logics are not only useful for representing common sense reasoning but also provide appropriate foundations for some scientific theories, and in particular for linguistics. Without a nonmonotonic formalism, it is very difficult to see the way to axiomatize a realistic planning domain, because the ways in which an effect may fail to be achieved or in which a background condition may fail to continue to hold after the action has been performed are practically impossible to enumerate.