| Abstract | The conventional view of speech perception holds that the sounds of speech are treated first as ordinary auditory objects and then matched to speech templates. An alternative view is one which states explicitly that speech perception is performed by a specialized system, often called a module, which treats speech sounds in the context of the vocal tracts that produce them. This view places humans squarely in the line of evolution, since it appears that every higher species has a neurological specialization for communication with its own species. The present paper reviews direct evidence that speech perception is specialized, seen in the phenomenon of duplex perception, in which the speech signal is artificially modified to such and extent that part of it is heard both as speech and as nonspeech simultaneously. Most commonly, this is done by placing a small part of the signal on one ear while the other ear receives the remainder. The speech process is strong enough to overcome this ecologically impossible situation, and it integrates the information from the two sources. The existence of duplex perception, then, immediately shows that speech perception is not dependent on auditory scene analysis, which assigns sounds to sources and sources to locations. Further exploration shows that the discriminability of the two percepts differs radically, indicating that two separable processes are independently at work. |
|---|
