Three experiments used sine-wave replicas of speech sounds to explore
some differences between speech perception and general auditory percep
tion. The experiments compared patterns of behavior in categorization
and discrimination tasks for listeners reporting either speech or nons
peech percepts of sine-wave replicas of speech. We hypothesized that t
he perception of speech sounds is automatized, while the perception of
less familiar sounds is not. The first experiment was designed to inv
estigate the perception of relatively long initial consonant transitio
ns using a synthetic /wa/-/ya/ sine-wave analog continuum. Speech list
eners perceived the continuum categorically, but nonspeech listeners c
ould not consistently categorize the items in the continuum. In the se
cond experiment, both speech and nonspeech listeners could consistentl
y categorize stimuli having final glides (an /ay/-/aw/ sine-wave repli
ca continuum), but differences between speech and nonspeech listeners
were found in the slopes of the identification functions, in reaction
times, and in the effect of context. These differences are consistent
with the hypothesis that speech perception is automatized. In the thir
d experiment, nonspeech listeners' discrimination sensitivity was grea
ter than speech listeners'. The observed pattern of results suggests t
hat speech perception is accomplished by a fast, obligatory, and thus
automatic perceptual mechanism.