Peter C. Gordon
Dept. of Psych., Univ. of North Carolina, Chapel Hill, NC 27599-3270
The coherence of vowels as auditory objects was studied by comparing identification thresholds in noise for synthetic vowel sounds (differing only in the center frequency of a single formant) to identification thresholds for the distinctive formant presented in isolation. The bandwidth of the noise masker was limited so that it only interfered with perception of the distinctive formant. Thresholds for accurately identifying the vowel sounds were lower than those for identifying the isolated formant. This demonstrates that vowel sounds cohere in the sense that unmasked formants reduce the masking of a formant embedded in noise. The advantage of a complete vowel over an isolated formant appears to depend on the temporal alignment of the formants. When the onset of the distinctive formant coincides with the offset of the other formants, then listeners can still identify the vowel sound in modest amounts of noise. However, in this case thresholds are not lower for vowel identifications than for identifications of isolated formants. This indicates that temporal synchrony plays a basic role in the psychoacoustic coherence of vowels.