A foray into the lexical semantics of distributivity

Lelia Glass (Stanford)


Some predicates are inherently distributive on their subject argument, such as laugh: when we know that Ana and Bob laughed, we know that Ana laughed, and that Bob laughed. This sort of distributivity inference is usually tied to lexical semantics of laugh (Scha 1981, Link 1983, Roberts 1987, Champollion 2014, de Vries 2015). But which verbal predicates are inherently distributive in this way, which ones are not, and why? As a preliminary step towards systematically investigating the lexical semantics of verbal distributivity, this talk observes two puzzles.

Looking at the first puzzle, we observe that VPs built from intransitive verbs (like laugh) are overwhelmingly distributive on their subject argument, whereas VPs built from transitive verbs (build a raft) are much more likely to have a collective interpretation available (Ana and Bob built a raft - together ). Moreover, among transitive VPs, it is unclear which ones will have a collective interpretation available and which ones will not. All of these patterns call for explanation.

Turning to the second puzzle, we observe that (a)telicity and (non)distributivity are not related in a straightforward way, in that {telic, atelic} and {collective, distributive, both} predicates can be found in any combination; however, we also observe that atelic VPs often favor exclusively distributive interpretations (the atelic predicate Ana and Bob ate applesauce can only be distributive), whereas telic VPs often favor optionally collective interpretations (the telic predicate Ana and Bob ate an apple can be distributive or collective).

These puzzles are deeply related (Champollion 2010), in that both telicity and distributivity engage the question of how much participation in a VP event counts as having VP’d yourself? In other words, for which VPs can Ana participate in an "Ana and Bob VP" event without the VP being separately true of Ana? As a partial solution to these puzzles, I propose that VPs describing a scalar change realized on an object can have both a distributive and a collective interpretation available because it is possible for one atom (e.g. Ana) of a multipart subject (Ana and Bob) to help to move the object along the scale without fully doing so independently, allowing for an interpretation where Ana and Bob brought about the event together but not separately. In contrast, intransitive VPs do not involve subjects realizing a change on an object; instead, it is the subject that is directly undergoing the change described by the VP, so there is no way for a member of the subject (such as Ana) to help to bring about the event without directly participating in it enough for the VP to be separately true of her.