Neural Control of Segmental Timing
George D. Allen · The Journal of the Acoustical Society of America · 1973
This paper attempts to specify precisely what kind of neural mechanism underlies segmental timing in speech production. One possible mechanism involves a neural “clock,” with a very short cycle time (of the order of 1 or 2 msec), which is read by a neural counting buffer. The duration of a speech segment could then be coded neurally as some number, n, of clock cycles, and timing control would be realized by the counting buffer counting up to n. A second possible mechanism involves a neural “decay” process (e.g., decreasing rate of firing of some neuron) and a neural “zero detector” to tell when the process has decayed. Segmental duration could then be coded as the starting value given to this decay process. No neural mechanism is error free, however, and the errors produced by these two hypothetical mechanisms should be different. The clock and counter model, since it involves a large number of clock cycles per segment, predicts symmetric error structure; the decay model, on the other hand, predicts positively skewed errors. Data to test these predictions are presently being collected. [This investigation was supported by PHS Research Grant No. DE 02668 from the National Institute of Dental Research and in part by General Research Support Grant No. RR 5333 from the General Research Support Branch of the National Institutes of Health.]