Between “cost” and “default”: a new approach to Scalar Implicature

Scalar Implicatures are pragmatic inferences that are normally derived in conversational exchanges when a scalar term, such as for example “or”, is used. Different theoretical accounts have been proposed to describe how and at which point in the derivation we actually add this inference. Large part of the most recent debate is focused on the question of the “cost” of implicature computation, an aspect that is crucial to choose among alternative accounts. In this perspective, my intent here is to present an experimental study in the ongoing debate centred on the “costly” or “default” nature of implicature computation. The main result of the study presented here is the fact that a “cost” is found only when the implicature is added despite the fact that it leads to a weakening of the overall assertion (namely, in DE contexts): this loss in informativity, and not implicature computation per se, is interpreted as the source of this “cost”. The theoretical background for this study is offered by Chierchia (2006) and his new intriguing parallelism between the phenomenon of scalar implicature and negative polarity.