On Friday, March 1, Vincent Homer (École normale supérieure) will give a colloquium entitled "Polarity and Grammar." The talk is at 3:30 in Machmer E-37. An abstract follows.
The licensing of polarity items is an old and difficult problem. It is hard to find the exact conditions that govern the distribution of those elements (e.g. some, any), not simply due to the intrinsic complexity of the patterns, but also because it is not even clear whether one should look for conditions couched in syntactic or semantic terms.
The discovery, due to the seminal work of Fauconnier (1975) and Ladusaw (1979), that downward-monotonicity is a characteristic property of the set of expressions whose presence can make an NPI acceptable (e.g. any in (1)), was a breakthrough which could have decisively tipped the scales in favor of a semantic approach.
(1) John didn’t understand anything.
That is, it could have become universally accepted that an NPI is only acceptable in a sentence S if it is in a position within S in which downward inferences are supported. But in fact, a vast majority of researchers who work on the topic view licensing essentially as a syntactic relation between a polarity item and an operator equipped with a certain negative feature, and do not take the monotonicity of the environment of the NPI as a direct factor. There is at least one good reason that supports their move: adding a negative expression to (1) does not lead to anti-licensing (2), as might be expected on strictly semantic grounds, given the monotonicity reversal; the acceptability of (2) is expected on the other hand if all that any requires is a structural relation with at least one appropriate operator above it:
(2) It is impossible that John didn’t understand anything.
In this talk, I propose a general account of licensing, both for NPIs and PPIs, which goes against the grain of the dominant syntactic approaches, as it maintains that polarity items are directly sensitive to the monotonicity of their environment. The theory, which relies on a few simple principles, also accounts for a number of unnoticed facts, and sheds new light on the dedicated mechanisms which evaluate the acceptability of polarity items, and on their interaction with other interpretative processes.
