Plural Marking and Inherently Plural Predicates
(joint work with Marcelo Ferreira)

Abstract:

It is currently assumed that phi-features, and in particular plural features, are interpreted on DPs and uninterpreted on Vs. Correlatively, the features on Vs would be born as 'unvalued', their value being assigned via a valuation relation traditionally called 'agreement'. Cases of paradoxical agreement have been solved by assuming regular, non paradoxical agreement with invisible features that would sit at a higher abstract level inside the DP. In this paper we will present data from Brazilian Portuguese that cannot be explained against this background. We will instead need to assume that (i) the Number feature on Vs can be valued as singular by default (rather than via agreement with/copying of the Num feature of the DP subject and (ii) the Number feature on V constrains the interpretation of V: singular valuation blocks 'inherently' plural readings. Our proposal will be shown to shed new light on Basque data. The constraint in (ii) holds for both Brazilian Portuguese and Basque, distinguishing between 'inherently plural' and 'cumulative' predicates. Since in both cases the subject refers to an entity that has more than one element, this semantic distinction cannot be assumed to be inherited from the subject DP. This will lead us to discuss various issues that arise regarding the semantics of plural-marking inside DPs on the one hand, and on inflected verbs, on the other hand. On the syntactic side, we will be concerned with the relation between agreement/valuation and interpretability of features.