The Perception and Cognition of Racialized Voices

Human listeners are able to determine racial and/or ethnic background from a voice based on acoustic and linguistic cues alone (Perrachione et al, 2010; Reaser et al, 2004). However, the mechanisms that underlie this determination remain poorly understood. Here we report a series of three experiments that each contribute to understanding what properties of the voice guide a listener to decide whether a talker is black or white. A speech corpus of continuous, read sentences is used in each experiment as experimental stimuli. The first experiment involves a norming study conducted online whereby subjects (n = 200) listened to each sample and completed two-alternative forced choice (2AFC) tests to identify the race of the speaker. The stimuli identified correctly above chance level were then used in the second experiment, where we conduct a series of comparisons between the acoustic qualities of black and white speakers. Finally, the third experiment involves changing the salient cues identified in experiment 2. After applying manipulations to the audio, we test whether or not listener categorization changes in another two sets of online studies. Results indicate that the ability of listeners to identify speaker race is affected after manipulating acoustic properties of duration, F0, and the harmonics of the sound. These features are likely just some of the cues that influence a listener's determination of race from the voice. Taken together, we observe that physical qualities in a voice factors into discerning speaker race. This study has implications for understanding and reducing racial prejudice.