S. argues that there is no decent pragmatic methodology. For one thing, there is disagreement already about what is part of semantic content and what is not. Further, there are not sufficient tests to find out whether an implicature is conversational or not. Moreover, of Grice's characteristics of conversational implicature, 'nondetachability' is not necessary, 'cancellability' maybe is, and only 'calculability' clearly is necessary, but then in a trivial sense.