Title: Social Signal and discourse function

Extracting information and structuring information from recordings
multimodal and interactional events, such as speeches, meetings, etc.
appears to be an open-ended problem. The challenge of these contexts
resides in the presence of interaction and social dynamics between
participants: hesitations, interruptions, overlaps, laughter and
pauses may constitute noise for traditional information extraction
techniques. On the other hand, those features of interactional
contexts may well be source of information. Social signals, for
example, are a fundamental component of conversational interaction and
their role is crucial in understanding social dynamics in multiparty
communication. However, in addition to contributing to social dynamics,
they may also have a function in the discourse structure.

In this work we focus on laughter, we focus on laughter, exploring
whether laughter can signal structural development of conversation,
such as topic change. We investigate the relation between laughter
and topic changes from two different points of view (temporal
distribution and content distribution). We also explore the impact
of participants’ acquaintance. Results reproduce effects evident in
earlier studies in terms of the proximity of laughter to topic
changes and refine previous conclusions about the differences
in topic change signalling between shared and solo laughter (the
difference appears to be partially a function of acquaintance).
We conclude that laughter has quantifiable discourse functions
alongside social signalling capacity.


Bio: Francesca Bonin is a 2nd year PhD Student in the School of
Computer Science and Statistics, under the supervision of Carl Vogel
and Nick Campbell. She holds a MSc. in Language Technologies and a
B.A. in Digital Humanities from the University of Pisa (Italy).
Before starting her PhD in Trinity College, she has worked on
different research projects between the Institue of Computational
Linguistics of the National Research Council of Pisa and the Language,
Interaction and Computation Laboratory at the University of
Trento (Italy), working on information extraction and terminology
extraction.