Removing the Disguise: The Matched Guise Technique, Incongruity, and Listener Awareness

Abstract

Sociophonetic perception is often studied using versions of the matched guise technique (MGT). Linguists using this technique appear united in the methodological assumptions that participants believe the manipulation and that this belief influences perception below the level of introspective awareness. We report an audiovisual matched guise experiment with a novel “unhidden” instruction condition. The basic task is a replication of the Strand effect (Strand 1999; Strand and Johnson 1996). Participants in the “unhidden” condition were instructed that the man or woman in the photo did not represent the voice they were listening to. Participants in both guises exhibited the Strand effect to nearly numerically identical extents. This result suggests that participants need not believe a link exists between a voice and a purported social category for visually cued social information to influence segmental perception. We explore the implications of this result for the MGT and for theories of social awareness and speech perception more broadly.