DISCUSSION PAPER

The interesting papers we heard all dealt in one way or another with a question that is surely central to a n inquiry into the biology of language: Are linguistic processes in some sense special, different from the processes that underlie nonlinguistic activities, and, perhaps, unique to man? To discuss that question, and the papers of the evening’s session, I find it useful to distinguish two classes of specialized processes: auditory and phonetic. Specialized auditory processes would serve, perhaps in the fashion of feature detectors, to extract those aspects of the acoustic signal that carry the important information. One is led to suppose that such devices might exist because it is true, and paradoxical. that some of the most important phonetic information is contained in parts of the speech sound that are not physically salient. Thus, a significant acoustic cue is in the formant transitions, though these are often of short duration and rapidly changing frequency. Perhaps there are devices devoted to detecting those transitions. If so, we should hold them up as examples of specializations in the auditory system. They would be important for the perception of language, but not properly part of its special processes. If the acoustic signal were directly related to the phonetic message, then detection of the phonetically important cues would be sufficient for phonetic perception; no further processing would be necessary. But the relation between signal and message is peculiarly complex. (For summary accounts, see Refs. 1-5.) As a result, the specialized auditory detectors can only begin the job; the auditory display they produce must still be interpreted, because the phonetic message is there in such highly encoded form. If there are devices specialized to d o that kind of interpreting, then 1 should consider them phonetic, not auditory. Since 1 will organize my comments on the papers of the evening in terms of that distinction, I should take a moment to illustrate what 1 mean. Consider the formant transitions that are important cues for the perception of stop consonants in syllable-initial position, and call up in your mind‘s eye spectrographic representations (similar to those shown by Dr. Morse) of such transition cues as would be appropriate for [da] and [ba]. Now add a patch of fricative noise-the hiss of [sl-just before the [da]. If that patch is immediately in front of the [da], you will hear [sa], not [da]; the stop will have disappeared completely. But if the patch is moved away so as to leave about 50 msec of silence between the end of the hiss and the beginning’of the formant transitions, then you will hear [sta]; that is, you will hear the stop once again. The generalization that captures those facts, and many others closely related to them, is that a necessary condition for the perception of syllable-initial stop consonants is a brief period of silence in front of the appropriate transition cues. But why should silence be necessary? Why should it be impossible to hear the stop when its acoustic cues follow closely on the fricative noise? The simplest explanation, surely, is that we are here dealing with a characteristic of the generalized mammalian auditory system. That might seem reasonable if only because in putting the fricative noise in front of the transition cues we have