The Situated Multimodal Facets of Human Communication - Coverbal Synchrony in Human-Machine Interaction - page 184

for male) and medium length (81% for female, 33% for male children)
are followed by changes of scene that signal both the production
of a greater amount of added information and boundaries in the
discourse structure. Only a low percentage of short pauses (20%
for females and 8% for males) serve this purpose, which probably
reflects the role of pausing due in cognitive efforts to recall from
memory and lexicalize concepts not yet known by the listener. It can
be hypothesized that more cognitive effort is involved in providing
more added information (a change in the scene requires a change in
the discourse topic), resulting in longer pausing times. The data also
suggest a predictive scheme for the alternating pattern of cognitive
rhythm in the production of spontaneous narratives. In this alternating
pattern, long pauses account for the highest percentage of paragraphs
or changes followed by medium pauses. Even though more frequent
than medium and long pauses, there is a low likelihood that short
pauses may signal a change of scene (20% in female and 8% in male
children) in the narration flow.
As in adults (Chafe, 1987), children use empty pauses of short,
medium and long duration to cue both the added information they are
providing to the listener and the cognitive effort they are making to
recover and lexicalize it. Pausing strategies act as cues indicating to the
listener the novelty of the received information and its cognitive cost,
and allow him/her to carry out the appropriate inferences. Pausing
strategies are also used as a linguistic tool for discourse segmentation,
marking word, clause and paragraph boundaries. Previous works
(Esposito et al., 2004; Esposito, 2005, 2006) showed that 56% of child
pauses indicate a major transition in the speech fl ow helping to plan
the message content for the continuation of the discourse, suggesting a
common pausing strategy for structuring the discourse. In addition, the
distribution of pauses of short, medium and long length is consistently
adopted among the subjects suggesting a coarse pause-timing model
that speakers exploit to regulate speech fl ow and discourse organization.
It would be interesting to make sense of how this model works by
collecting more data. Nevertheless, the results would be of great utility
in the fi eld of human-machine interaction, favoring the implementation
of more natural speech synthesis and interactive dialog systems.
4. The Role of Gestures
We are not conscious of gesturing and I was extremely surprised
when my colleagues in the United States, in mimicking my speech,
were exuberantly moving their arms and hands. Then, I learned that,