[PS-1.89] What would a compositional hearer do? - controlling for prior expectations in visual world timecourse studies

Sun, C. & Breheny, R.

University College London

Much visual world research has examined whether comprehenders interpret some as some and not all in the same timecourse as they compute the semantic interpretation of all. Here we show that previous studies have not fully controlled for prior expectations relating quantifier use and visual information and we demonstrate a new way of controlling for such factors by visually separating the target girl with some/all of the socks from a residue set of objects that determines which girl has some or all prior to the onset of socks.

Offline tasks (Experiments 1-2) demonstrate low-level expectations for the all referent to be the big set and the some referent to be the small set. Experiments 3-4 adapt previous visual world studies, making it possible to examine anticipatory looks to the residue. We compare both pragmatic some and all conditions to number conditions (girl with two/three of the socks). The latter items do not require looks to a residue set to verify the target. We find greater looks to residue for both quantifier conditions compared to numbers and no difference between. We found that prior expectations influence eye movements to the target making the interpretation of previous visual world data problematic.