Colloquium with Tanya Bondarenko, Harvard University

Title: In favor of equality semantics: an argument from Russian polarity subjunctives

Abstract: Theories of clausal embedding according to which embedded clauses denote predicates of events or individuals with propositional content (Kratzer 2006, Moulton 2009, Bogal-Allbritten 2016, Elliott 2017 a.m.o.) differ in what the relationship established between the content event/individual and the embedded proposition is. Subset semantics proposes that the output of the Content function is a subset of the embedded proposition (Kratzer 2006, 2016, Hacquard 2006, a.o.), whereas  Equality semantics equates the two sets (Moulton 2009, Elliott 2017, Bassi & Bondarenko 2021, Bondarenko & Elliott 2026 a.o.):

(1) a. Subset Semantics: λx. Content(x) ⊆ p

    b. Equality Semantics: λx. Content(x) = p

In this talk I'll examine polarity subjunctives in Russian, and argue that they provide an argument in favor of the Equality semantics for clausal embedding. I will show that there is a class of clause-embedding predicates in Russian with which subjunctive clauses exhibit behavior of weak NPIs: they are available only in Strawson-Downward Entailing contexts. This phenomenon raises two questions: (i) how are these subjunctive complements licensed?; (ii) why do we see this behavior of subjunctives only with the verbs that we see it with? I will argue that what unites the relevant class of verbs is that the meanings of their embedded clauses lack displacement: they denote sets of exemplifying situations rather than sets of individuals with propositional content. I'll make a proposal within the alternatives-based approach to mood (e.g. Villalta 2006, 2008) about how such subjunctives are licensed, and show that the Equality semantics of clausal embedding, unlike Subset semantics, correctly predicts that in the general case, clauses that denote predicates of events/individuals with content should fail to license polarity subjunctives.