On the Conversational Basis of Some Presuppositions

The current literature on presupposition focuses almost exclusively on the projection problem: the question of how and why the presuppositions of atomic clauses are projected to complex sentences which embed them. Very little attention has been paid to the question of how and why these presuppositions arise at all. As Kay (1992, p.335) observes, “treatments of the presupposition inheritance problem almost never deal with the reasons that individual words and constructions give rise, in the first place, to the particular presuppositions that they do.”1 This is the question on which this paper will focus.

There are two kinds of answer that one might give to the question of how presuppositions arise. One type of answer is that presuppositions are conventional properties of lexical items, as in the conventional implicature view of Karttunen and Peters (1979). On this view, certain lexical items have, in addition to their truth conditional content, a special presuppositional content, which is carried through the compositional process to produce a propositional presupposition. Although the Karttunen and Peters model for treating presupposition has been rejected by most current researchers, our talk about presupposition seems at least implicitly to take their view of the sources of presuppositions for granted: we talk about the presuppositions of know, of too, and so on, as if assuming that the presuppositions are properties of these items.2

Presuppositions, however, might be thought to have a very different source. Presuppositions might be conversationally derived, that is, they might be inferences which are licensed by general conversational principles, in combination with the truth conditions of the presupposing utterance. Stalnaker, from whom we have inherited the currently standard view of presupposition, suggests repeatedly that at least some presuppositions have a conversational source. Indeed, he sees one of the primary advantages of the move from a semantic to a pragmatic account of presupposition as being the possibility of explaining “some of the [presupposition] facts in terms of general assumptions about rational strategy in situations where people exchange information or conduct argument” (1974, p. 205). However, Stalnaker never attempts to work out any general derivation for conversational presuppositions, nor any means for distinguishing conventionally generated presuppositions from those with a conversational source.

What we have established in this discussion is that at least some presuppositions must have a conversational basis, that is, they must be inferences derivable from some general conversational principle or principles. I would like to conclude by considering what consequences this conclusion has for the general treatment of presupposition.

On the view of presupposition now most standard in the literature, presuppositions are thought to be propositions which must be entailed by the presumed common ground of the discourse participants. However, if at least some presuppositions are derived by the kind of mechanism which gives rise to other conversational inferences (i.e. conversational implicatures), then it is more appropriate to view them as propositions which the addressee can infer the speaker to believe on the basis of what the speaker has said, plus the assumption that the speaker is behaving cooperatively.16 On this picture, the derivation of a presupposition may require speaker and addressee to share certain assumptions (e.g. that the speaker is behaving cooperatively), but the presuppositions themselves are neither required nor expected to be entailed by the common ground.

Moreover, on this picture, presuppositions are not attached to atomic clauses, but are inferences derivable from the utterance as a whole, given the conversational situation. This raises a question about algorithmic treatments of presupposition projection, which are predicated on the assumption that presuppositions are locally generated.