Abstract
Within the Gricean framework only what is conversationally implicated is cancellable, whereas what is conventionally implicated and what is said cannot be cancelled without giving rise to contradiction. In the relevance-theoretic framework, however, the question is whether explicatures, which replace the Gricean notion of what is said, are cancellable. In recent years, various objections to the cancellability of explicatures have been raised. The aim of the present paper is to demonstrate that these objections are due to a misinterpretation of the Gricean cancellability test. In particular, they disregard the fact that this test is merely one of several diagnostic tools that are used by Grice to distinguish between conventional and conversational implicatures. Once we have recognized the essence of the cancellability test, the objections to the cancellability of explicatures turn out to be unwarranted.