Antipresuppositions

In A. Ueyama(ed.), Theoretical and Empirical Studies of
Reference and Anaphora: Toward the establishment of generative grammar as an empirical science (2006)
by Percus, Orin

Abstract

Some sentences come with an interesting kind of condition on their use: they require that a certain proposition p not be taken for granted. Recent work by Sauerland and Schlenker proposes an explanation: the sentences in question compete with other sentences that require that p be taken for granted, and win only when those other sentences are infelicitous. This paper is concerned with the precise nature of this competition, and with what can be learned from it.

Ugh... from our "Toward the establishment of generative grammar as an empirical science" Dept.

This thing here called Anti-presupposition is a decent observation about some interesting types of clauses in English. My problem is with the author's attributing some of these observations to what ultimately amounts to our language module.

One such case that results from the analysis is the "empirical evidence" that the quantifier 'all' is infelicitous if applied to a set that maximally contains two elements. So a sentence such as (1) sounds bad (is ungrammatical?) and is dispreferred to (2)

1. *The mafia broke all of his legs
2. The mafia broke both of his legs

This 'anti-duality' condition, which results from anti-presupposition triggered by the quantifier, is a result of the fact that English happens to have a special quantifier for exactly these cases ('both'). I think this is a fine hypothesis, which appears to be verified by its extension across other languages that have a 'both'-like quantifier... until you look at French, which seems to show this 'anti-duality' but doesn't have a quantifier that is 'both'-like enough. But I digress...

Time will tell if this hypothesis will live on. But my problem is with the underlying notion that this anti-duality condition, like the anti-presupposition notion, is entirely language bound and not a result of general cognitive constraints (or simple frequency issues). A lot of effort is spent covering all the instances of anti-presupposition and anti-duality, but underlying this is the unquestioned assumption that this phenomenon is a product of UG. This fault is going to affect our ability to, as the abstract suggests is the goal of the article, learn something from anti-presuppositions.