To join the meeting, please click https://uva-live.zoom.us/j/85868588169.

Abstract

Downward entailing (DE) quantifiers are more difficult to process than upward entailing quantifiers (UE) (e.g., Deschamps, Agmon, Loewenstein, & Grodzinsky, 2015). When verifying sentences with DE quantifiers, participants produce more errors and have longer reaction times. Two theories proposed different explanations of these effects. According to the two-step processing model, the DE quantifiers are represented as corresponding UE quantifiers with a hidden negation. For example, “fewer than half” is equivalent to “NOT (more than half)”. The longer reaction times are hypothesized to reflect an extra step related to the processing of hidden negation (Clark, 1976; Just & Carpenter, 1971). The pragmatic account, in contrast, claims that DE quantifiers decrease the predictability of sentence continuation (Nieuwland, 2016) and that they are dispreferred in most of the contexts. The experimental and modeling data provided evidence in favor of both accounts (Schlotterbeck, Ramotowska, van Maanen, & Szymanik, 2020).

In this talk, I will present an analysis of EEG data from an experiment involving two quantifiers: "more than half" (UE) and "fewer than half" (DE) (Augurzky, Schlotterbeck, & Ulrich 2020). I used Hidden semi-Markov Model multivariate pattern analyses (HsMM-MVPA, Anderson, Zhang, Borst, & Walsh, 2016) to detect the stages of processing of quantifier sentences in EEG signal. The two-step account predicts an extra stage of processing for DE quantifiers related to hidden negation. The pragmatic account predicts that the stage related to context-updating and decision making will be longer. The initial results do not provide evidence for the extra stage of processing of DE quantifiers and rather support the pragmatic account.