Jon Gajewski will give a CRISSP Lecture Series from December 19 to December 21. The title and the abstract for the lectures are now available: “Polarity and Truth Conditions.”
More information on the time, date, venue and registration can be found in the events section.
These talks will concern how the distribution of negative polarity items is affected by truth-conditional and non-truth conditional meaning. Von Fintel has influentially argued that one kind of non-truth conditional meaning, presupposition, must be factored out of the licensing conditions on polarity items. Chierchia, on the other hand, has shown that another kind of non-truth-conditional meaning, implicatures, can interfere in NPI-licensing. I will argue that both are correct, but that there are important additional generalizations to be made about when and where non-truth-conditional meaning matters for determining the acceptability of a polarity item.
Evidence for this view will come primarily from examination of the distinction between weak and strong NPIs. Weak NPIs are those like English ever that have been argued to appear in downward entailing environments, cf. Ladusaw’s work. Strong NPIs appear in a proper subset of the environments that weak NPIs appear in. Zwarts influentially proposed that the distribution of strong NPIs can be captured with the formal property of anti-additivity. I have argued for a different view that takes the different roles of non-/truth-conditional meaning into account. In this talk, I will argue that strong NPIs show greater sensitivity to non-truth-conditional meaning than weak NPIs.
One issue that will have to be clarified is what aspects of the environment of a polarity item are relevant to determining the acceptability of a polarity item. For example, one must decide if there is an operator that can be identified as the constituent whose semantic properties license the occurrence of a polarity operator or if the presence of a polarity item is sanctioned by the total semantic properties of a constituent that contains it (roughly licenser- vs. environment-based approaches). I will argue for a hybrid of the two approaches whereby both properties of the licenser and material between the licenser and polarity items must be considered. Having investigated the separate roles of truth-conditional meaning and non-truth-conditional meaning in licensing, I argue that the licenser must be treated differently from other material within the environment of licensing.
Pursuing this account will lead us into discussion of problematic cases. The first problem case is definite descriptions. Negative polarity items can in limited cases occur in definite descriptions. The new approach to NPIs that I advocate requires re-examining the distribution of NPIs in definites. The second class of problem cases involves complex quantificational expressions I predict to license strong NPIs but do not. I will argue that these are not true counterexamples on the grounds that these operators do not create licensing environments when viewed from a suitably spare perspective on logical form. In this regard, I will discuss previous work on unacceptability that derives from trivial truth conditions.