It is a little harder to say what the fundamental period is. For example, in this example it could be 400 ms, in which intervals of 200 ms would come about by insertion and intervals of 800 ms would arise from deletion. Both insertion and deletion are required to create intervals of 600 ms. Alternatively the basic period could be 200 ms, in which case all larger intervals arise by beat deletion. Or it could be 800 ms with a lot of beat insertion. You get the point. We can document the presence of a rhythm as a statistical tendency, without necessarily being able to write the musical score for the speech!
So what did you find?????
Basically, we found that when people read prose fluently there is often (but not always) evidence of an underlying period. Most interestingly, this tendency is much stronger when people are asked to read very quickly while still remaining intelligible. Some fast readings showed a surprisingly strong periodic component-comparable to that found in say a jazz recording.Below you can see the beats and amplitude envelope from a small section of each of two readings. In each case the original text spoke here is " ...section, accenting the same beat". The top figure shows beats from the slower reading. Notice that the beats are quite irreguarly spaced. In the faster reading, the beats become much more regularly spaced. Listen to the two readings (the recordings provide a little more text than is shown in the figures).
One way in which the beat series has been regularized is by a stress shift. In the slower reading the subject says ...SEC-tion AC-cent-ing the SAME beat while in the faster reading she says ...SEC-tion ac-CENT-ing the SAME beat, with a stress on the second syllable of accenting. Although this would be a funny place to stress the word if spoken in isolation, it seems to be ok here as it helps to establish the regular beat period.
Beats from a slower reading
Slower reading
Beats from a faster reading
Faster reading
So what?
This second experiment showed that if we conceive of speech rhythm as a succession of beats, we can indeed find cases where the beats line up neatly, and we can distinguish rhythmical readings from less regular ones. Furthermore, it looks as if speech is more rhythmical when people speak in a rapid and fluent manner than when they are free to pepper their speech with pauses and hesitations. No big surprise perhaps, but it is reassuring to be able to document the fact in an experiment.In the following page, the last of the present tour, the results of the two experiments presented here will be tied together and some speculation about their import for the study of action and coordination will be entertained.