The emergence of conventions in a turn-taking task

Gregory Mills

A central finding in dialogue research is that interlocutors rapidly converge on idiosyncratic referring expressions. Convergence is inherently interactive, relying on participants providing each other with both positive and negative evidence of understanding (Clark, 1996; Healey, 2007) and alignment (Garrod and Pickering, 2009).
In addition to reference, interlocutors also need to coordinate procedurally on how to take turns with each other. Dialogue is replete with procedural expressions that establish who performs which action, when it should be performed, and how initiation and completion of the action should be signaled, e.g. “wait a moment”, “when I’ve done x, do y”, “you first”, “while I do X do y”. Recent work (Knutsen, et al., 2018) has demonstrated that even in the canonical tangram joint reference task (Clark and Wilkes-Gibbs, 1986), 30% of all expressions are concerned with procedural coordination. Moreover, when such expressions are used interactively, they rapidly become conventionalized (Mills, 2014; Fusaroli et al., 2014). However, it is currently unclear how the central dialogue mechanisms of positive and negative feedback and alignment contribute towards conventionalization.
To investigate how procedural coordination develops, we report an “alien-language” task which prevents participants from using natural language by restricting typing to a limited set of keys. However, in contrast to canonical alien-language tasks which elicit referring expressions, this task elicits procedural expressions.
For example, on a typical trial, the task of the Director might be to instruct the Matcher that they have to perform the following 5-step sequence:
(1) Matcher presses A;
(2) Matcher presses S
(3) Director and Matcher press F simultaneously
(4) Director presses J at the same time as Matcher presses D
(5) Matcher presses D
This task presents participants with the recurring procedural coordination problem of communicating and then performing a wide variety of sequences of actions, without using natural language. In order to test the putative role of feedback on how coordination develops, dyads were assigned to one of 4 conditions:
(1) Positive feedback: participants can send Y for yes
(2) Negative feedback: participants can send N for no
(3) Positive and Negative feedback: participants can send Y and N
(4) No feedback: participants cannot send Y or N
Participants who could provide both positive and negative feedback solved more trials, confirming the grounding models basic predictions. Surprisingly, participants who could signal negative feedback performed worse than participants who were blocked from providing any feedback. We argue this pattern is due to the intrinsic ambiguity of “no”: it signals but does not diagnose coordination problems, giving no indication what the next relevant action should be. Moreover, participants aligned more in unsuccessful trials, contradicting the interactive alignment model (Garrod and Pickering, 2009). We argue this is due to participants using alignment as a repair strategy: if participants know the next action, they perform that action. However, when participants misunderstand their partner, a default strategy is to repeat their partners behaviour in order to establish a basic level of coordination that can then serve as the scaffolding for more complex communicative strategies.


Clark, H. H. (1996). Using language. Cambridge university press.
Fusaroli, R., Rączaszek-Leonardi, J., & Tylén, K. (2014). Dialog as interpersonal synergy. New Ideas in Psychology, 32, 147-157.
Garrod, S., & Pickering, M. J. (2009). Joint action, interactive alignment, and dialog. Topics in Cognitive Science, 1(2), 292-304.
Healey, P. G., Swoboda, N., Umata, I., & King, J. (2007). Graphical language games: Interactional constraints on representational form. Cognitive Science, 31(2), 285-309.
Knutsen, D., Bangerter, A., & Mayor, E. (2019). Procedural Coordination in the Matching Task. Collabra: Psychology, 5(1).
Mills, G. J. (2014). Dialogue in joint activity: complementarity, convergence and conventionalization. New ideas in psychology, 32, 158-173.