Yorùbá Speech Surrogacy with the Dùndún Talking Drum: Recognition Accuracy is a Function of Language Familiarity and Music Training

The Nigerian dùndún talking drum is known for functioning as both a speech and song surrogate for the Yorùbá language. Previous research confirmed the dùndún can accurately mimic temporal, fundamental frequency (f₀), and intensity characteristics of Yorùbá vocalizations when used as a speech surrogate, and that this acoustic match systematically decreases for drumming modes in which more musical context is specified. Here we expand on that work with a two-by-two cross-cultural behavioral study comparing identification accuracy of Nigerian Yorùbá versus American English speakers, and musicians versus nonmusicians. Thirty or more participants in each two-by-two category (total N = 124) compared 45 seven-second drum excerpts to matching or non-matching speech and song excerpts and rated perceptual similarity on a 1–10 scale. Results statistically confirmed that similarity ratings for matching pairs are an additive function of both language familiarity (Yorùbá > English) and music training (Musicians > Nonmusicians), with language familiarity accounting for more variance. We also found Drum-VocalSong excerpt pairs produced higher similarity ratings than Drum-VocalSpeech, consistent with greater information density in speech messages. Our findings verify the importance of individual differences and confirm the independent contribution of language familiarity and music training on the effectiveness and potential functionality of speech surrogacy communication systems.