Notes on presupposition and order of composition 1

The standard judgment (which I share) is that (1a) asserts rather than presupposes that France has a king. In contrast, (1b) presupposes that France has a king. Assuming that the and in (1) is the simple bivalent logical connective, the key difference appears to be the linear order of the conjuncts: in (1b), the information that France has a king arrives too late to help satisfy the presupposition triggered by the definite description in the first conjunct. In dynamic semantic accounts of presupposition in the tradition of Heim (1983), the linear asymmetry is built into lexical items corresponding to the various logical connectives. But as Soames, Schlenker, and others point out, making these assumptions on a per-lexeme basis missing the systematic pattern that the asymmetry always favors earlier material over later material. So it is necessary to state the linear asymmetry independently of specific lexical items. But on the standard conception of the division of work between syntax and semantics, semantic composition depends entirely on function argument structure, and not on linear order. It may not even be clear how semantic interpretation could be sensitive to linear order. Schlenker concludes that the behavior of presuppositions must crucially be stated in terms of non-semantic objects such as strings or syntactic trees.