As a bonus to the guest blogs the on IPrA conference, I’m delighted that Elliott Hoey has accepted an invitation to contribute a report on Tanya Stivers’ keynote address. This was one of the most thoughtful and original presentations, raising as it did the question: did the evolution of talk favour two, and only two, people?

The taskmasters at IPrA convinced a sizeable crowd to attend an 8:30am plenary talk by Tanya Stivers, who suggested that conversation is “built” for dyadic interaction. Her abstract can be found here.  The following is not verbatim reporting, but simply what I recall.

She began by noting that two participants are minimally required for talk-in-interaction; that three present the possibility of ‘marginalizing’ a third party; and that four or more allow for schisming. Primae facie evidence for a two-party bias comes from studies on crowds. These show that when people are allowed to move about and interact freely, what emerges is a preponderance of dyadic formations, relatively few triads, and diminishing numbers of 4-party and 5-party interactions.

Select only one next speaker

For conversational interaction, she pointed out that current-selects-next techniques normally select but one speaker, which is seen in how gaze, pronouns, and pointing often converge to single out just one person.* On top of this, the organization of sequences appears to favor a

single SPP. If more than one party could conceivably answer a question, then participants treat the provision of just one answer as adequate.

And indeed, she showed a case where the provision of a second SPP (from a different speaker) was oriented to as non-normative. Finally, she noted that there’s a physical constraint on gaze direction: you can only gaze at one person at a time, such that the rapidly changing facing-formations result in a series of exchanges that are best analyzed as dyadic.

Operating against the bias

She ended her talk on an optimistic note, saying that the two-party tilt of conversation should spur investigations into how multiparty talk is achieved. Though there appears to be a bias toward dyads, that is, interactional participants routinely work to include other parties—for instance, through the practices associated with laughter and storytelling.

What is ‘the structural organisation of conversation’?

I had been looking forward to Stivers’ talk ever since reading the abstract, and overall, I found it stimulating, plausible, and provocative. To my mind, it can be seen as a kind of call to recenter of our analytic focus. As the reach and influence of CA extends increasingly beyond the orderly realm of talk-in-interaction (e.g., Haddington et al., 2014; Nevile, 2015), her keynote is a welcome reminder for us to pin down precisely what we mean when we talk about the structural organization of conversation.

---

• Footnote: She held up recent work by Kendrick and Holler on ‘response queueing’ (2015) as a contrastive but infrequent case. Responses may be ‘queued’ through gaze, such that for certain questions in which two answers are reasonably anticipated from two question-recipients, the participant who is gazed at at the TRP tends to respond first, followed by the other recipient.

References

Haddington, P., Keisanen, T., Mondada, L., and Nevile, M., Eds. (2014). Muliactivity in Social Interaction. Amsterdam: John Benjamins.

Kendrick, K. and Holler, J. (2015). Response queueing in multiperson interaction. Paper presented at 14^th IPrA, University of Antwerp, Belgium.

Nevile, M. (2015). The embodied turn in Research on Language and Social Interaction. Research on Language and Social Interaction, 48(2), 121-151.