Abstract
One of the most well-established sound-shape correspondences is that the majority of people pair the word “Bouba” with rounded patterns while matching “Kiki” with angular patterns. To date, this correspondence has primarily been demonstrated using arbitrary visual patterns, and hence very little is known regarding the precise visual characteristics underlying this correspondence. In the present study, we examined the crossmodal mapping between bouba/kiki and radial frequency patterns in order to systematically manipulate the visual features of a pattern. A radial frequency pattern, a closed-contour with sinusoidal modulation along its radius, is widely used in shape perception studies and considered as mid-level visual processing. Three dimensions of the radial frequency patterns were manipulated: shape frequency, shape amplitude, and spikiness (by increasing the number of harmonics of triangular waveforms added on the top of rounded contour). In each trial, the participants (N = 74) viewed a radial frequency pattern on the screen and judged that either “bouba” or “kiki” (presented auditorily) provided a better match. The results demonstrate that all three factors modulated the sound-shape correspondence interactively. When the shape frequency increased from 3 to 11 cycles/360 degrees, the associated sound shifted from “bouba” to “kiki”. In the intermediated shape frequency ranging between 5 and 7 cycles/360 degrees, the associated sound shifted from “bouba” to “kiki” either when the shape amplitude or the spikiness was increased. These results suggest that shape frequency dominated over the shape amplitude and spikiness in determining the sound-shape correspondence that was observed, implying that visual global/local processing may underpin this particular sound-shape correspondence. We suggest that radial frequency patterns thus provide an intriguing visual stimulus set with which to systematically investigate the crossmodal correspondence between bouba/kiki and rounded/angular shapes. Furthermore, a possible hierarchy of crossmodal correspondences including early-level features, mid-level patterns, and high-level object can be established.
Meeting abstract presented at VSS 2015