James S. Magnuson, Reiko Yamada, Howard C. Nusbaum
The Effects of
Talker Variability and Familiarity on
Mora Perception and Talker Identification
Abstract:Contextual tuning theories of talker normalization state that listeners can use information about a talker's vocal characteristics stored in working memory to facilitate recognition of that talker's speech [9]. We investigated whether people can use information about a familiar talker's voice, stored in long-term memory [12, 10], in the same way. That is, can subjects circumvent normalization processes when listening to highly familiar talkers, such as family members? We found that familiarity with a talker's voice facilitated
performance in a talker identification task, and also in a mora identification task with degraded stimuli. However, in a monitoring paradigm that typically results in faster performance in single-talker than multiple-talker conditions[9], we found the typical normalization effect for both familiar and unfamiliar talkers. Thus, while information about talkers that listeners have in long-term memory can be used strategically to facilitate, e.g., segment identification, that information is not available until the initial processes of talker normalization are complete. We discuss the implications of the results for theories of talker normalization and talker identification.