Phonetic imitation of the acoustic realization of stress in Spanish: production and perception

This study explores imitation of the acoustic realization of Spanish stress in disyllabic words produced in isolation, which is cued by three correlates: F0, duration, and intensity. Forty-eight native speakers of Mexican Spanish shadowed one of four model talkers of the same dialect. Differentials for each acoustic correlate of stress were generated by calculating the difference between the values of the first and second vowels for each of F0, duration, and intensity, for all recordings. Next, 87 Spanish speakers participated as listeners in a holistic perceptual assessment (4IAX task) of the shadowers’ productions. Bayesian mixed-effects modelling was performed for both the acoustic and perceptual data. The results showed that the shadowers imitated the model talkers on all three differentials, but made the greatest shifts on the F0 differential, followed by duration, shifting the least on intensity. Analysis of the perceptual pattern showed that the listeners perceived imitation and that the shadowers’ imitation on all three differentials contributed to the perceptual pattern. Lastly, the extent to which the listeners relied on imitation of the differentials aligned roughly, but not exactly, with how much the shadowers had converged on each differential, with listeners using imitation on duration the most, followed by F0, followed by intensity.