This question is closely tied to the notion of esoteric vs. exoteric communication (Thurston 1987; Wray & Grace 2007). The hypothesis is that esoteric communication is common in small groups with high amounts of interpersonal communication and shared norms. These communication patterns are hypothesised to result in certain kinds of interaction styles and linguistic patterns. In contrast, exoteric communication is thought to be common in larger groups where people will likely interact with complete strangers, which will beget interaction and linguistic patterns that are different to those of esoteric communication. These broad communication types are hypothesised as impacting the evolution of different linguistic structures. Dyad vs group communication is one way to characterise this distinction in communication types.
The scale is as following:
Communicative pressures of dyadic vs group-like interactions are different, and have consequences over time for how a communicative system changes (cf. Atkinson et al. 2018). Dyadic vs group-like interaction is also found to affect how individuals are influenced by others. Experimental work by Fay et al. (2000) found that dialogue-like interaction show dyad members being influenced most by those with whom they interact in the discussion, while group-based communication is more monologue like, members are influenced most by the dominant speaker. This would have consequences for modelling the diffusion and adoption of linguistic innovations through a social network.
What exactly constitutes a dyad-like vs group like configuration varies across studies. For example different experiments choose different thresholds. The Fay et al. (2000) experiment had up to 5 people for the dyadic condition, and over 10 as the group condition. Atkinson et al. (2018) did their experiments on pairs vs triads to test for emergence of communicative conventions. Fay & Ellison (2013) have a pairs vs group of eight set up for their iterated learning experiment investigating the effect of population size.
We have chosen to follow the Fay et al. (2000) set up, defining dyad-like as a group of up to five people, while groups of over five are considered more “group like” (see Fay et al. (2000) for empirically grounded rationale). We have chosen this number since real-life interactions of people are rarely completely dyadic, so we surmised that a number of people which still produced a dyad-like effect is sufficient to capture the intended effect of the esoteric communicative condition.