From Seeing to Saying: Perceiving, Planning, Producing

148 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2009.Given the amount of visual information in a scene, how do speakers determine what to talk about first? One hypothesis is that speakers start talking about what has attentional priority, while another is that speakers first extract the scene gist, using the obtained relational information to generate a rudimentary sentence plan before retrieving individual words. These five experiments evaluated these views by examining the conditions under which different types of information may be relevant for production. In Experiment 1, participants told the time from analog clocks. Eye movements revealed that it was not the visual features or configuration of the clock hands, but rather the upcoming referring time expression that determined gaze patterns. Differences in eye-voice spans also suggested a process in which scene elements are relationally structured before a linguistic plan is executed. Experiments 2--5 employed a modified version of Gleitman, January, Nappa, and Trueswell's (2007) attentional cuing paradigm. Participants were more likely to begin picture descriptions with a particular actor if their attention had been drawn to it. Eye-voice spans demonstrated that participants started speaking before they had planned an utterance framework. However, these effects were modulated by the amount of time given to extract the scene gist (Experiments 3--4) and by the ease of identifying the pictured event and actor names (codability norming and Experiment 5). This suggested that perceptual factors influence word ordering only when conceptual information is not immediately available or insufficient for generating an utterance framework.U of I OnlyRestricted to the U of I community idenfinitely during batch ingest of legacy ETD