Mandatory implicatures in Gricean pragmatics

Gricean implicatures are often viewed as a very weak kind of implication, viz., as optional enrichments of the literal meaning of an utterance. In virtue of being optional, they need not be present on every occasion of utterance, and they can be explicitly denied. Further, if implicatures are viewed as optional enrichments, a false implicature cannot be the reason why an utterance is infelicitous: If an implicature is known to be false, it should simply fail to arise. I argue that this view is mistaken. There is no reason to think that Gricean reasoning cannot lead to very robust inferences, which are neither optional nor cancelable, and whose falsity can render an utterance infelicitous. Moreover, I point out that a number of explicit recent theories do not exclude such robust implicatures without additional stipulation, and that informal appeal to Gricean reasoning triggering infelicity has been made in the investigation of a number of apparently semantic phenomena. I conclude that implicatures are only weak, optional and cancelable if the pragmatic pressures driving them are context-dependent in a particular way, and further that there is no reason to think that all pressures are of this kind.