[PS-3.6] Appearances aren't everything (Or are they?) Exploring effects of clipart versus photograph stimuli on measures of on-line language processing

Saryazdi, R. & Chambers, C.

University of Toronto

Many psycholinguistic studies employ either photographic or clipart images as stimuli, with convenience/availability typically serving as the only reason for choosing a particular image type. However, developmental work has shown sensitivities to images that vary in terms of abstraction, and adult studies have shown effects of perceptual detail (colouring/texture) on aspects of visual cognition. Here we explore whether image type (clipart/photograph) affects measures of on-line language processing. We first developed a new stimulus set (~200 object images) for use in visual-world experiments by digitally transforming photographs into closely-matched clipart analogues. Listeners viewed arrays of four images, accompanied by a recorded sentence ('Jamie will peel/move the banana'). Image type was manipulated across blocks. Experiment 1 (verb-based anticipation) showed more efficient visual processing with photograph stimuli (inspection of more scene areas, shorter fixation durations) before the sentence began. Some residue of this effect was apparent at the verb, with initially faster identification of verb-congruent referents with photograph stimuli. Experiment 2 (semantic category activation) showed that images of category associates (e.g., APPLE, for the noun in 'Jamie will move the banana') again reflected greater visual consideration when they were photographs. Results are described in terms of implications for visual-world studies of language processing.