Perception of gender in children's voices

Abstract

Research shows that gender in the voices of prepubescent children can be identified relatively well, despite the absence of reliable anatomical differences related to their vocal tract morphology. This implies that information about differences in girls’ and boys’ voices also resides in behavioral and cultural aspects of speech and not exclusively in the acoustic cues corresponding to anatomical differences. The observed improved gender identification (gID) for sentences over syllables suggests that longer stretches of speech provide listeners with a richer acoustic basis for their decisions. Here, we introduce two additional factors that may affect gID: speaking style and children’s dialect. Children (n = 92) ages 8;0–12;4 years, boys (n = 45) and girls (n = 47) from Ohio, Wisconsin, and North Carolina produced isolated syllables, read sentences, and spontaneous utterances. Listeners were young adults from Ohio. Expectedly, gID accuracy for sentences/utterances was greater (66%) than for syllables (55%), and for older than younger children. Dialect was a significant predictor for younger (but not older) children, with the highest gID for Ohio. Overall, spontaneous utterances did not provide more gender information than read sentences except for older children from North Carolina, suggesting that cultural traits in conversational dynamics can also be useful to listeners in gID.