---

Gwyneth Lewis, Dominique Crocitto, Megan Zirnstein
University of Memphis

Subject Listing - Psychology
Advisor: Dr. Max Louwerse

Thursday, Oral Session 3, Presentation 5, Carmichael Hall 102

ARE PICTURES WORTH AN INFINITE NUMBER OF WORDS?

This study investigated the ways visual information can be inferred from spontaneous dialog that is in turn based on visual information. We predicted that dialogue is interpreted similarly, but that the inferred information is not very accurate when expressed in visual information. To test these predictions, subjects were asked to read transcribed dialogs from a previous study involving the Map Task scenario and to draw subjective versions of the maps involved. The study occurred in two separate phases. In the first phase, three expert raters completed maps for the dialogs. In the second phase, untrained subjects were asked to recreate two maps each from one selected dialog. The measures used were intended to elicit how accurate the recreated maps were in comparison to not only each other, but also a gold standard. Dependent variables included token number, token location, token location relative to the route, and overall route length. The recreated maps were found to share many similarities on all measures with each other as well as with the original maps from which the dialogs were based. Considering how similar the recreated maps were to each other and to the original maps, this study suggests that visual information is similarly inferred and accurately conveyed through linguistic information.

Advisor: Dr. Max Louwerse, Assistant Professor, Dept. of Psychology/Inst. for Intelligent Systems, University of Memphis, Memphis, TN