Perception of voice qualities in implicit and explicit phonetic convergence

Talkers' speech becomes more similar to the speech of others through phonetic convergence. While the effects of phonetic convergence are well-documented for a number of acoustic and perceptual dimensions, the present study examined convergence to voice quality in a word shadowing task. This study additionally sought to understand differences in production when participants are asked to repeat the words they hear (implicit shadowing) as compared to when participants are explicitly asked to imitate speech. Convergence to voice quality was assessed acoustically through measurement of f0, the difference in amplitude between first and second harmonics (H1–H2), and cepstral peak prominence (CPP), and perceptually through an AXB task in which participants assessed the perceptual similarity of shadowing productions to the productions of model talkers. In general, both these analyses of participants' speech showed phonetic convergence toward model talkers' voice qualities. Preliminary results suggest the degree of convergence was characterized by the instructions in the shadowing task. Explicit instructions generally resulted in greater convergence but also resulted in convergence to different acoustic features as compared to implicit instructions. Complementary to studies describing quantitative differences in the magnitude of convergence across these conditions, these data also evidence qualitative differences in which acoustic dimensions are preferenced in convergence.