P-Centers

(Note: the American spelling for 'centre' is normally used).

The P-Center is one of the most interesting and perplexing discoveries in speech science. The original publication on the subject was a brief paper by J. Morton, S. Marcus and C. Frankish in Psychological Review, 1976, vol 83.5, pp. 405-8. They reported that in trying to produce test tapes of regularly spaced English numbers, sequences in which the time intervals were exactly equal tended to sound unequally spaced. It's quite difficult to hear this, but here's an informal demonstration: in the first recording, I recorded the numbers 1 to 10 in what I thought was a regularly-spaced way. Then I modified the intervals between the numbers so that the acoustic onsets (the moment when the sound started) were equally spaced. See if you can hear any difference in the timing.

Recording 1

Recording 2

This preliminary paper was followed by one by Steve Marcus in Perception and Psychophysics, 1981, vol. 30, pp. 247-256 entitled 'Acoustic Determinants of P-Center Location.

Clearly, this effect is quite powerful, and must have an effect on speech perception. This was followed up by Hilary Buxton (1983) "Temporal predictability in the perception of English speech". In: Prosody: Models and Measurements. Ed. by A. Cutler and D. R. Ladd, Springer-Verlag, Berlin, pp. 111-121. This showed that items spaced according to P-Center timing were more accurately perceived than items spaced at acoustically equal intervals.

These finding have an important consequence: studies of speech timing have to take account of P-Centers if they are to say anything meaningful about speech perception. Looking for "stress-timing", for example, in terms of exact measurement of the intervals that can be observed in the acoustic signal do not reveal what determines our perception of temporal regularity. This is explained clearly by Elizabeth Couper-Kuhlen  English speech rhythm : form and function in everyday verbal interaction, Amsterdam : J. Benjamins, 1993.
.