Transparency ( An Incremental Theory of Presupposition Projection ) 1

The analysis of presupposition projection lead researchers to posit in the early 1980's that the meaning of a clause should be viewed as its Context Change Potential rather than as its truth conditions (Heim 1983, following Stalnaker 1974). We argue that this 'dynamic turn' was misguided, and that it leads straight into a dilemma: either one follows Stalnaker in his pragmatic analysis, in which case one obtains a beautiful analysis of presupposition projection in conjunctions, but not of much else; or one follows Heim in her semantic analysis, which yields broader empirical coverage but little explanatory depth (no predictions are made about connectives whose Context Change Potential was not stipulated to begin with). We sketch an alternative account, entirely developed within classical logic. We argue that in some cases a complex meaning m is conceptualized as involving a precondition p, with m=pp' (Division) In this case a pragmatic principle, Be Articulate!, requires that if possible m should be expressed as a conjunction p and pp' rather than as pp' (in order to make explicit the special status of the precondition p). If so, why can pp' ever be pronounced on its own? Because a principle of Minimization sometimes rules out the full conjunction p and pp', leaving pp' as the sole contender. Specifically, Minimization prohibits any sentence S from starting with α[p and if it can be ascertained that no matter how S will end the string p and could be deleted from S without modifying its contextual meaning. This derives a principle of Transparency, which suffices to obtain Heim's projection results, but which has the advantage of making predictions for connectives that she does not consider. We end the paper with some speculations on the Principle of Division, and suggest (but do not show) that it is itself pragmatic in nature. 1 The Dynamic Turn and the Transparency Theory 1.1 The Dynamic Turn and an Alternative A new conception of meaning arose in the early 1980's, thanks in large part to the works of R. Stalnaker and I. Heim. It used to be thought that the meaning of a clause could, as a first approximation, be analyzed as its truth conditions. Building on ideas by Stalnaker, Heim suggested instead that one should view the meaning of a clause in a dynamic way, as the effect it has on the Context Set, i.e. on the beliefs of the speech act …