Seminaire SynSem

Sentences with plural definites allow what have been labeled as "non-maximal readings," i.e., readings that allow exceptions. To illustrate, a sentence like (1) is clearly a true depiction of a situation where all the circles are red, but it can also sometimes be accepted as a true depiction of a situation where some, but not all, of the circles are red. The availability of these readings has been shown to depend on the context of the conversation (Malamud 2012).

(1) The circles are red.

Schwarz (2013) found experimental evidence from response times that non-maximal readings take longer to process than maximal readings, which is taken as an indication of a higher processing cost. However, so far, there has been no experimental work investigating the underlying cognitive processes behind the derivation of non-maximal readings. I will present a new experiment that seeks to address two questions:

(i) Are non-maximal readings more cognitively costly than maximal readings?
(ii) Does the derivation of non-maximal readings involve a two-step cognitive process?

The experiment consisted of a truth-value judgment task, during which several measures were collected. Question (i) was assessed by comparing response times for non-maximal and maximal responses, with the goal of conceptually replicating Schwarz’s findings. To answer question (ii), mouse trajectories from participants were collected. The underlying hypothesis is that a two-step derivation will manifest as a more deviated trajectory in the non-maximal trials than in the maximal trials.