dc.description.abstract | Conversation is a collaborative activity. In face-to-face interactions interlocutors have mutual
access to a shared space. This thesis aims to explore the shared space as a resource for coordinating
conversation. As well demonstrated in studies of two-person conversations, interlocutors
can coordinate their speech and non-verbal behaviour in ways that manage the unfolding conversation.
However, when scaling up from two people to three people interacting, the coordination
challenges that the interlocutors face increase. In particular speakers must manage multiple listeners.
This thesis examines the use of interlocutors’ bodies in shared space to coordinate their
multiparty dialogue.
The approach exploits corpora of motion captured triadic interactions. The thesis first explores
how interlocutors coordinate their speech and non-verbal behaviour. Inter-person relationships
are examined and compared with artificially created triples who did not interact. Results demonstrate
that interlocutors avoid speaking and gesturing over each other, but tend to nod together.
Evidence is presented that the two recipients of an utterance have different patterns of head and
hand movement, and that some of the regularities of movement are correlated with the task structure.
The empirical section concludes by uncovering a class of coordination events, termed simultaneous
engagement events, that are unique to multiparty dialogue. They are constructed using
combinations of speaker head orientation and gesture orientation. The events coordinate multiple
recipients of the dialogue and potentially arise as a result of the greater coordination challenges
that interlocutors face. They are marked in requiring a mutually accessible shared space in order
to be considered an effective interactional cue.
The thesis provides quantitative evidence that interlocutors’ head and hand movements are
organised by their dialogue state and the task responsibilities that the bear. It is argued that a
shared interaction space becomes a more important interactional resource when conversations
scale up to three people. | en_US |