Skip to main content

News & Events


Jon Ander Mendia
Cornell University

Friday, April 30, 2021


This event has been canceled.


Pragmatic weakening and the proviso problem

Presuppositions of compound sentences do not follow a homogeneous projection pattern. Notably, presuppositions triggered in the consequent of a conditional or the second disjunct of a disjunction are varyingly inherited by the complex sentence: either it is inherited wholesale (like "Theo has a wet suit" in (1a)), or a weaker “conditionalized” inference is generated (like "If Theo is a scuba diver, then he has a wet suit" in (1b)).

(1) a. If Theo is in a generous mood, then he will bring his wet suit.
      b. If Theo is a scuba diver, then he will bring his wet suit.

A prominent response to this problem–the “proviso problem” (Geurts 1996)–is to assume that in addition to a basic weak conditional presupposition, there is a pragmatic strengthening mechanism that allows for unconditional inferences (e.g. Beaver 2001; von Fintel 2008). But such approaches have recently been criticized as under-predicting unconditional presuppositions (Mandelkern 2016a,b). In this talk I explore an analysis of variable projection from complex sentences that takes the opposite view. Refining a Gazdar-style (1979) Cumulative Hypothesis (cf. van der Sandt 1988), I suggest that presuppositions fail to project because of general conversational principles: presuppositions project globally unless they “cause pragmatic embarrassment” (Beaver & Geurts 2011). I submit that one such case of pragmatic embarrassment is that where speakers, by virtue of admitting that a certain presupposition is known, declare that their epistemic state is inconsistent. The gist of the idea is the following: if speaker S is ignorant about proposition p, a complex sentence will not be taken to presuppose p, since, if it did, the speaker would have to be assumed to hold an inconsistent epistemic state. I focus here conditional and disjunctive statements, environments where we can easily identify the effect of inconsistent epistemic states in presupposition projection.
This way of thinking about presupposition projection highlights the distinction between the presuppositions of an expression and what it takes for a context to satisfy them (Karttunen 1974). I discuss how relying on this form of "admittance conditions" we can handle Mandelkern's (2016a,b) as well as other well-known problematic cases.