The effect of talker familiarity on stream segregation

This study contrasts different forms of familiarity with a talker’s voice, to better explore how these types of familiarity might influence a listener’s ability to understand that voice in the context of noise. Listeners were asked to shadow a target voice while a second voice spoke fluently in the background. Listeners differed in their familiarity with the target voice: one group was familiar with the voice and were told explicitly whose voice they would be hearing (explicit knowledge); a second group was familiar with the target voice but were not warned whose voice it was (implicit knowledge), and members of a third group were entirely unfamiliar with the target voice. Explicit knowledge of talker identity appeared to have the larger effect on listener performance: Participants with explicit knowledge made significantly fewer shadowing errors than those with only implicit familiarity. Familiarity did influence the types of errors listeners made: those participants who were familiar with the target voice prior to the test session made fewer incorrect responses than did those who had not heard the talker previously, although their total number of errors (including misses, as well as incorrect responses) did not differ. In a second experiment, neither explicit knowledge of a distracter voice’s identity nor implicit familiarity with the background speaker had any effect on shadowing a target voice. This suggests that familiarity with a voice only helps listeners when that voice is the one being attended. These studies suggest that there may be more than one manner in which talker-specific information influences perception. r 2005 Elsevier Ltd. All rights reserved.