# On the relation of speech to language **Abstract:** phonetic unit to which it is somehow linked, because, as should be said again, the initial representation is already phonetic. To see the phonetic module even more clearly as a component of the larger language module, a theorist does well to note how far its processes resemble those of syntax. Consider, then, how the phonetic module presumably goes about deciding that some signal does or does not contain information about phonetic structures. The important point is that it can hardly do that by reference to the surface properties of the acoustic signal, as might be seen most readily in the case of sine-wave speech. There, the formants and the bandlimited noises (of the fricatives) are replaced by sinusoids that follow the centers of the formants and the noises. Those sinusoids have no common fundamental, no common movement, nor indeed, any acoustic property that would make them coherent from an auditory standpoint. Yet, if they describe phonetically telling trajectories of the articulators, listeners hear phonetic structures39,40. Thus, the only basis for perceptual coherence is at the same time gestural and phonetic. That being so, we should not expect to identify surface acoustic characteristics that will reliably distinguish three sine waves that cause the listener to perceive phonetic structures from three acoustically similar sine waves that will be heard as unrelated tones. In that respect, phonetic 189 Trends in Cognitive Sciences – Vol. 4, No. 5, May 2000