Spectral cues influence perception of voice onset time when using the /ɑ/ vowel: Pitfalls and solutions

Perception of voice onset time (VOT) has frequently been used to test auditory temporal processing. However, the first formant frequency (F1) of an adjacent vowel systematically covaries with VOT, presenting a confounding spectral cue. The most problematic confound between VOT and F1 would be for a vowel such as /ɑ/, the most commonly used vowel context for VOT studies in the literature. The present study tested the hypotheses that (1) F1 cues influence perception of VOT when followed by the low vowel /ɑ/ but not when followed by the high vowel /i/, and (2) this confounding spectral cue in /ɑ/ is so salient that it is perceptible for cochlear implant users. Stimuli included a /d/-/t/ VOT continuum coupled with /ɑ/ or /i/, where VOT-F1 covariance was controlled through the ratio of VOT duration to the duration of the vowel that was replaced by aspiration (i.e., cutback ratio). For both listener groups, VOT-F1 covariance affected perception of consonants followed by /ɑ/ but not those followed by /i/. Studies pertaining to auditory temporal processing that have used consonant + /ɑ/ stimuli should be reevaluated because of the likely spectral confound. Future work in this area should use the /i/ context to more effectively isolate temporal cues.