This study investigated the perception of English consonants that show variation in the U.S. South: interdental fricatives /θ ð/ produced with different degrees of stopping and lateral liquid /l/ produced with different degrees of vocalization. The question of whether the categorization of these consonants is racially biased was examined in a task, where participants rated sounds by clicking along a visual analog scale in two conditions: the audio only condition (word recordings) and the audio-visual condition (recordings paired with pictures of White and Black speakers). The scale represented the “th” – “t", “th” – “d", or “l” – “no l” continuum. A hypothesis that all listeners would perceive interdental fricatives as more stopped and the lateral liquid as more vocalized when paired with Black speaker pictures in the audio-visual condition was not supported. /θ/ and /l/ were perceived less as intended sounds (i.e., interdental and lateral liquid) when produced by Black speakers than by White speakers. Listeners' perceptions of /ð/ did not differ greatly by the listener or speaker's racial background. The results suggest an interaction between a dialectally salient sound target and speaker or listener race that is not influenced by additional visual speaker clues.
Morales et al. (Tue,) studied this question.