Threshold comparisons for intensity-based duplex perception for speech

Abstract

Duplex perception (DP) occurs when a stimulus component simultaneously contributes to two percepts (Rand, 1974). A proposed DP variant involves substituting a tone glide for an F3-transition in a /da/ or /ga/ syllable. At full intensity the glide is perceived as a nonspeech chirp that simultaneously distinguishes the consonant. It also has been claimed that the glide contributes to speech (consonant) perception over a range of intensities that are insufficient for nonspeech (chirp) detection, raising the possibility that a phonetic module takes precedence over nonspeech perception (Whalen and Liberman, 1987). Others have failed to find a range of precedence when chirp detection thresholds were reevaluated using a 2AFC task (Bailey and Herrmann, 1993). Which conclusion is valid? The current investigation addressed this question by comparing individual thresholds for consonant perception and chirp detection using a variety of methods. Chirp detection thresholds were evaluated using the method of constant stimuli, 2AFC, and various reminder tasks (using a fixed standard). Thresholds for consonant perception were evaluated using identification and discrimination (same-different) tasks. Results analyzed according to both traditional threshold and detection theories support the claims for a range of precedence and suggest that discrepant chirp detection thresholds reflect a reliance on consonant perception.