Nonspeech-nonspeech auditory contrast

Abstract

Many perceptual adjustments for a target speech sound's context have two explanations: listeners could compensate for coarticulation or perceive the target as contrasting auditorily with its context. Only auditory contrast can explain listeners' similar adjustments for non-speech contexts that acoustically resemble the original speech contexts. This experiment tests auditory contrast further by replacing the target sound with non-speech, too. Stimuli consisted of a sequence of two equal ERB fraction spaced tone complexes separated by a short gap. Either the same higher or lower 1 ERB-wide band of tones was amplified in both complexes, creating HH and LL stimuli, or a band 1 ERB higher or lower, creating LH and HL stimuli. Tones were amplified to create HL, LH, HH, and LL stimuli near lower, intermediate, and higher frequencies (1800, 2300, and 2800 Hz). If frequency change within a stimulus evokes auditory contrast, then HL-LH stimulus pairs should be more discriminable than HH-LL pairs, where frequency only changes between stimuli. Listeners discriminated HL-LH stimulus pairs better than HH-LL pairs in all three frequency ranges, as predicted if the complexes contrast within but not between stimuli. They also discriminated both kinds of pairs worse at higher than lower frequencies.