Sound symbolism is the non-arbitrary association of sound and meaning. Experiment 1 demonstrates that sound symbolic associations facilitate the online processing of male and female voices when the target words contain the vowels /a/ and /i/ for male and female voices, respectively, when listeners are engaged in a speeded gender identification task. Experiment 2 reveals that when listeners are attending to a vowel identification task, there is no effect of talker gender. This suggests support of models of perception where activation of linguistic structure masks indexical bottom-up information.

Index Terms: speech perception, speech processing, sound symbolism, adaptive resonance theory