Grammaticalization of Category Mistake and Natural Language

Abstract

The paper discusses the notion of a category mistake in natural languages. We argue that the fact, that sentences containing category errors are not marked as ungrammatical in natural languages, is their advantage and not a flaw. Argu­ments built on this notion are often used by philosophers against their opponents (the most well-known examples are G. Ryle’s criticism of dualism and R. Car­nap’s criticism of Heidegger’s metaphysical views). Carnap famously argued that sentences containing category errors are meaningless and that the fact that natural languages allow such sentences is their serious flaw. He argued that in a logically perfect language such sentences would be marked as ungrammati­cal. Some contemporary systems adopt a similar perspective. The paper argues that this position makes epistemologically undesirable predictions and is prob­lematic from the semantic point of view. From the epistemological perspective, the nature and properties of objects that can be referents of various terms in a natural language might not be known in advance, thus whether some state­ment contains a category mistake can only be established empirically by testing a hypothesis that must be first formulated. From the semantic perspective, a sen­tence containing context sensitive terms can be a category mistake in one context and not be such in another. The authors also explore cases of the so-called struc­tural tautologies and contradictions that are ruled as ungrammatical in natural language and argue that they are evidence for the sensitivity of natural language to the logical form, contrary to Carnap.