The effects of talker familiarity on talker normalization

Despite the tremendous amount of phonetic variability in speech across talkers, listeners seem to effortlessly process acoustic signals while attending to both the linguistic content and talker-specific information. Previous studies have explained this phenomenon by providing evidence for talker normalization, a process in which our perceptual system strips away information about a talker so that the abstract, canonical linguistic units are all that remain for further linguistic analysis. However, it is currently unknown whether or how talker normalization is facilitated by familiar talkers. In this study, we investigated whether talker familiarity had an impact on the speed in which listeners perceived highly confusable words under varying contexts. Over the course of three days, listeners were explicitly trained on the voices of four talkers. Baseline and post-test measures were administered to determine the effect of talker training and to see whether this effect was impacted by the presence of a carrier phrase as well as the variability of talker presentation. The results demonstrated that listeners adapted to the talker regardless of familiarity. Having immediate information about a talker from preceding speech appeared to play a larger role in managing talker variability than a long-term familiarity with the talker’s voice. Our findings suggest that talker normalization is a feedforward process that does not rely on prior memory traces