Ignorance in context: The interaction of modified numerals and QUDs

Matthijs Westera Adrian Brasoveanu ILLC, University of Amsterdam UC Santa Cruz I. The phenomena. Geurts & Nouwen (2007) report a contrast between superlative modifiers (henceforth Sups) like at most, and comparative modifiers (henceforth Comps) like less than: only sups license an ignorance inference. For instance, (1a) but not (1b) conveys that the speaker doesn’t know the exact number of diamonds: 1. a. Sup: I found at most ten of the diamonds under the bed. not sure how many b. Comp: I found less than ten of the diamonds under the bed. 6 not sure how many This contrast was experimentally confirmed by Geurts et al. (2010), but disconfirmed by Coppock & Brochhagen (2013), who use a different experimental setup. This paper builds on their work in search of an explanation for this discrepancy. We report the results of 2 experiments, each consisting of an acceptability-judgment-like task paired with a self-paced reading task, investigating the dependence of such ignorance inferences on questions under discussion (QUDs). Experiment 1 considered (1a,b) as responses to the 3 kinds of questions in (2) below (2× 3 = 6 conditions total):