• Le 18 octobre 2018
    De 14:30 à 17:00
    Campus Tertre
    Bâtiment Censive, Salle du LLING, C228

Jeudi 18 octobre 2018, 14h30
Responsable : Jérémy Kuhn (Institut Jean Nicod)

Titre: The dynamics of negative concord

Résumé:  Negative polarity items appear (roughly) in downward entailing environments. Why? Perhaps because in these environments, widening the domain strengthens the utterance (Kadmon & Landman 1993); perhaps because exhaustifying alternatives in these environments doesn't result in contradiction (Chierchia 2013); perhaps because in these environments one scopal ordering entails the other (Barker 2018). All these explanations turn out to pick out (roughly) the same contexts.
Negative concord items appear in a smaller set of contexts: roughly, those that are anti-additive (Zwarts 1998) or anti-veridical (Giannakidou 2000). Why? Here, semantic explanations are scarcer (but see Chierchia 2013). But here's one semantic property that these environments have: they prevent discourse referents from being introduced. If (1) is true, then the set of discourse referents (DRs) introduced by `nobody' --- the people that were seen by Mary --- is empty.
(1) Maria non ha vista nessuno.
Mary not has seen nobody.
`Mary hasn't seen anybody.' (Italian)

In this paper, I propose that this is in fact, *the* explanatory property of NC items. NC items are indefinites that flag the fact (in their lexical semantics) that they will fail to introduce a discourse referent. After spelling this intuition out formally using dynamic semantics, we show that it has number of explanatory advantages: 
It correctly predicts that NC items must appear under a local anti-veridical operator.
If the presupposition that the DR set in empty is made at-issue, we predict the existence of negative uses of NC items: exactly what's attested in fragment answers and non-strict  concord languages. It perfectly unites negative concord with recent analyses of other concord phenomena: distributive concord (Henderson 2014) and definite concord (Bumford 2017).