Exploiting the Ganong effect to probe for phonetic uncertainty resulting from hearing loss

Abstract

The “Ganong effect” is the tendency to perceive an ambiguous speech sound as a phoneme that would complete a real word, rather than completing a nonsense/fake word. For example, a sound that could be heard as either /g/ or /k/ is perceived as /g/ when followed by “ift” but perceived as /k/ when followed by “iss.” Because the target speech sound (/g/ or /k/) is the same across both environments, this effect demonstrates the influence of top-down processing rather than simply bottom-up processing in speech perception. We hypothesized that degradations in the auditory system (including simulations of hearing loss or cochlear implantation), speech stimuli will be rendered more ambiguous, and an increased Ganong effect should be observed. Participants heard three speech continua that varied by spectral cues of varying speeds, including fast (stop formant transitions), medium (fricative spectra), and slow (vowel formants). Stimuli were presented with clear spectral quality, or with varying amounts of spectral degradation using a noise vocoder. Responses were analyzed within stimulus pairs using binomial logistic regression, revealing increased Ganong effect with degraded speech. This test is proposed as a simple measure of how phonetic uncertainty can lead to an individual’s increased reliance on top-down processing.