Felicity and Presupposition Triggers

Many current pragmatic approaches to presupposition, beginning at least with Stalnaker (1974), are based on a notion of felicity. Presuppositions are requirements a sentence places upon a context for an utterance of that sentence to be felicitous. These requirements in turn are induced by the presence of certain presupposition triggers, such as lexical items like factive verbs, or constructions like clefts. In this framework, the notion of felicity is left unanalyzed, relying upon speakers’ abilities to provide felicity judgments. Likewise the mechanisms by which presupposition triggers lead to requirements on context are left largely unexplored (as most attention has been focused on the study of presupposition projection). In this paper, I shall begin to explore the notion of felicity and how it relates to presupposition triggers. My discussion will center around the observation that the kind of felicity relevant to presupposition is not an entirely uniform phenomenon. I shall argue this by a closer examination of infelicity. Two important subcategories of infelicity will emerge, distinguished by their discourse status. Generally, infelicity induces an opportunity for conversational repair. In some cases, we will see, the repair is obligatory, while in others, it is merely optional. I shall argue that this distinction correlates to a division among presupposition triggers. One category, which we may call strong presupposition, leads to obligatory repair upon failure. The other, which we may call weak presupposition, leads only to repair optional status. I shall then attempt to explain the distinction between strong and weak presuppositions, by offering an analysis of presupposition triggers in a framework of dynamic intensional logic. I shall argue that a systematic treatment of a wide range of triggers can be given in terms of a single dynamic operator. I shall also show how the dynamic properties of this operator can help to explain the different discourse properties of presuppositions. In dynamic logics, two sorts of failure can be distinguished. Update instructions can fail by resulting in a fail state; but instructions can also fail by being undefined for a given input. I shall propose that weak presuppositions result in the former kind of failure, while