Simple auditory elements induce perception of a phonetic feature

Abstract

In this presentation, I demonstrate that certain nonspeech sounds can have perceptual phonetic value. I focus on a single phonetic/articulatory feature, lip rounding, and its detection in simple auditory stimuli. Behavioral experiments show that a rounding percept is possible for two types of nonspeech. One stimulus type, which yields the more robust response, is a complex periodic source filtered by a single narrow band reminiscent of a speech formant. The resulting nonspeech varies in perceived roundedness depending on the filter’s frequency, corresponding roughly with F2. The other stimulus type is a pure tone modulated upward in frequency. Preliminary results suggest that rounding can indeed be perceived on these sounds, but only with specific modulation rates within a certain frequency range. These findings indicate that minimally simple auditory objects, including pure tones and filtered bands, can be sufficient to encode phonetic information. Additionally, these two types of cues diverge in their ability to trigger this percept: a filtered band works as a static spectral cue, whereas a pure tone requires spectrotemporal modulation. This observation is consistent with findings that there are auditory STRFs specifically sensitive to modulation and the theoretical perspective that auditory organization directly predicts the processing of speech.