Influence of suprasegmental features on perceived ethnicity of American politicians

How accurate are listeners at identifying the ethnicities of political figures from one-word samples? Do suprasegmental variables provide a basis for these judgments? Tokens of six lexical items were extracted from speeches by seven male political figures of different stated ethnic identities. In a Mechanical Turk experiment, 94 listeners heard each token twice, then responded to the multiplechoice question: “What is the ethnicity of this speaker?”. While listeners overall performed with an accuracy rate below chance, certain speakers and tokens were more consistently identified than others, both accurately and inaccurately. Analysis indicates suprasegmental features including Intensity Ratio, Harmonics to Noise Ratio, Jitter Average. Pitch Peak Ratio, and Syllable Duration Ratio, contribute to listener judgments of ethnicity.