French Face-to-Face Interaction: Repetition as a Multimodal Resource - Coverbal Synchrony in Human-Machine Interaction - page 162

In terms of temporal alignment, Gesture 5 starts 0.88 s after Gesture
1, yet, contrary to what we saw in example 1, the repeat is much longer
than the model (0.88 s vs. 0.36 s) and is also more complex as well.
Whereas Gesture 1 is only composed of a stroke because it is part of
a series of gestures which we will not describe here since they are not
relevant to the present study, Gesture 5 contains a preparation and a
retraction. If we consider the stroke only, then the repeat is shorter
than the model as it lasts 0.32 s.
One may consider that there is a redundancy between gesture
and verbal repetition in this example with no new information added.
However, the repetition plays a role in the message structure as it is
the global pattern which is repeated including words and gesture
and which has a function of backchannel. The whole extract is very
collaborative: speaker A produces some argumentation as to who will
take care of the baby and speaker B collaborates to the argumentation
first producing the backchannel signal “ouais” ( yeah ), then repeating
“tout le temps” ( all the time ) and repeating his own yeah again. His
whole utterance forms a complex backchannel with a function of
acknowledgement. At the prosodic level, the model and the repeat
are clearly distinct at least because of the location in the IP and the
discursive function of each one. A produces “tout le temps” in the end
of the IP with a major terminal rising contour (about 100 Hz) followed
by a high plateau while B produces “ouais tout le temps ouais” as a
single IP with a minor rise on “temps” (around 30 Hz). At this point,
the two speakers seem to be reaching the end of a conversational
sequence which could be the reason why their overall pitch is so low.
The configuration of this repetition exhibits a compressed span as it
is often the case in backchannels. Concerning the gesture repetition
(Gesture 5), although it is quite clear in the video that speaker B is
repeating speaker A's gesture, the timing in the verbal modality is
rather a repetition of B's speech by A. In fact, B's “restes” ( [you] stay )
begins 0.197 s before A's “reste” ( [I] stay ), so that when speaker A is
beginning to utter “reste”, he has enough acoustic material to know
what is being said by B. B's “restes” is much longer so that speech
rate is not similar for the two speakers. The lengthening on “restes”
by speaker B directly corresponds to the lengthening of his copied
gesture. However, according to the location of the word “reste” in
the two repetitions, their contour is not quite the same (see Figure
7). Whereas speaker B's contour is a low plateau at the end of an IP,
followed by another IP (“ouais”) A's “reste” is in the middle of the
IP that ends on “ici” expressed with a rise. The configuration of the
repetition that also functions as an acknowledgement is in accordance
with the previous verbal repetitions of this sequence.