(Non-)exhaustivity in embedded questions: Contextual, lexical and structural factors

Abstract

Wh-questions often allow for either a Mention-Some (MS) or a Mention-All (MA) answer/interpretation, but what licenses MS answers remains unclear. In this paper, we systematically investigate a set of linguistic (wh-word, matrix verb, modality/non-finiteness) and discourse/contextual (questioner goals) factors that give rise to MS reading. We present evidence from two experiments showing that MS readings of embedded questions are not only available across various forms of embedded question, but also in environments that have been claimed not to license MS readings: specifically, finite clause embedded who-questions (e.g., Mary knows who came to the party). Moreover, acceptability of both MS and MA is influenced by contextual information. These results thus call into question a strictly semantic approach towards the resolution and interpretation of wh-questions, and an approach that claims obligatory licensing of MS by modals, and at the same time, provide empirical support for theories that incorporate the role of a speaker’s context-dependent discourse goals.