|Institution(en)||Humboldt-Universität zu Berlin|
|Ort||4th floor, room 403 (Seminarraum)|
In this talk I report on the major results of my DFG project “Definiteness in articleless Slavic languages” and based on them, I propose a new situation-sensitive semantics for bare NPs in Slavic. Some background: According to the dominant neo-Carlsonian analyses of bare NPs (Chierchia 1998, Krifka 2003, Dayal 2004, Geist 2010), bare (count) nominals start out their semantic lives as properties and are shifted - as needed - to kinds (by nom), entities (by iota), or existential quantifiers (by ex). Various ways of taming this powerful (and apparently overgenerating) system have been proposed, typically by preferring certain shifts over others or banning some shifts altogether.
The empirical work my colleagues and I have conducted sheds doubt on the neo-Carlsonian approach to bare NPs. As an alternative, I propose that argumental bare NPs should be treated as referential expressions derived by a default choice-functional shift, with the choice function relativized to a situation (Šimík 2019). Bare NPs selected by choice functions relativized to situations introduced in the clause (as in existential or existential-like statements) are akin to narrow-scope existentials, while bare NPs relativized to (matrix) topic situations are akin to definite descriptions. No type-theoretic difference is assumed for the “definite” vs. “indefinite” bare NP contrast. The predictions of this approach, such as strict association of bare NP scope with its situation binder or possibility of intermediate scope, are yet to be experimentally tested. For the moment, I present experimental and corpus evidence that renders the neo-Carlsonian program problematic (while being compatible with the proposed analysis).
In Šimík & Demian (under revision, in preparation) we attempted to find evidence for the presence of uniqueness and maximality in referential bare NPs. By doing that, we hoped to tap into the hypothesized iota/sigma-shift applied to singular/plural bare NPs under suitable conditions. We failed. We did detect consistent (albeit weak) uniqueness and maximality inferences in German definite descriptions, but no such thing in Russian, Polish, or Czech bare NPs, despite multiple replication attempts. We conjecture that the iota/sigma-shift is unlikely for bare NPs (the results are compatible with Heim’s 2011 idea that bare NPs in articleless languages are always existential, but this proposal has its own problems). In Šimík & Burianová (to appear) we used corpus to find out whether referential properties of Czech bare NPs correlate with their position in the clause (as broadly assumed in the literature). We found a strong effect of initial position - in line with Geist (2010), clause-initial bare NPs are highly unlikely to be existential (“indefinite”). At the same time, we found no evidence for Diesing’s (1992) Mapping Hypothesis (which has had many followers in the literature on Slavic) - there is no correlation between pre-/postverbality and bare NP semantic type (related evidence exists for Dutch; van Bergen & de Swart 2009). I also present some additional informal analyses on Šimík & Burianová’s corpus, showing (i) that existential readings of bare NPs - singular or plural - are commonplace (contra Dayal’s 2004 expectations) and (ii) that the existential construal can in most cases be blamed on the existential nature of the predication (lending indirect support to my situation-sensitive approach to bare NP meaning).