The perceptual magnet effect in Australian English vowels

Recent research (Kuhl, 1991) has suggested that the internal structure of v owel categories is graded in terms of stimulus goodness. It has been propos ed that a best instance stimulus reflects a central point or prototype, whi ch effectively renders within-category members perceptually more similar. D iscrimination experiments suggest a nonlinear relationship between acoustic and perceptual space near category centers (Iverson & Kuhl, 1995b). This p henomenon has been described as the perceptual magnet effect. The present s tudy investigated the presence of the perceptual magnet effect in five Aust ralian vowel categories. Australian English speakers identified, rated, and discriminated between a pool of 32 vowel stimuli that varied in F1 and F2 values. The results from Experiments 1 and 2 showed that subjects were able to judge the quality and identity of each stimulus and that a general grad ing of stimulus quality was reported. This was not symmetrical, and the sub jects' responses varied considerably. In Experiment 3, closer control of th e methodology in the discrimination task and of contextual factors influenc ing the test materials was exercised. Despite this, evidence of the warping of perceptual space in discrimination data was not found. In general, thes e results do not provide support for the existence of the perceptual magnet effect, and explanations for this finding are discussed.