Conversations, regardless of their modality, are a vital part of our everyday life. In recent years, we have tried to evaluate verbal conversation behaviour (e.g., timing, speaking style, movement) as a means to more realistically diagnose how well or poorly someone can converse with or without assistance. Withdrawal and dominance are often referenced as two conversational strategies used when conversation is difficult (e.g., in noise, with a hearing loss). We here look for evidence of both strategies in the behaviours of four-person conversations recorded in a lab. Tetrads composed of two older adults with and two older adults without hearing loss carried out topic-driven conversations under six conditions: in quiet and two levels of background babble with and without wearing hearing aids for the two participants with hearing loss. To assess the degree of interaction, we analysed vocal activity over the course of each conversation for contributions from each member; the binary outcome ‘contribution’ was calculated using a proportional threshold of speaking time in a given time window. There was no clear pattern across noise conditions of partial withdrawal or dominance for any given participant. This general lack of previously reported real-world strategies may be a symptom of reactivity in laboratory conversations. There was generally a higher degree of interaction (i.e., more polylogues) in conversations when the interlocutors with hearing loss were aided, but the effects of noise and aiding on the degree of interaction in the conversations were dependent on the time-window and contribution-threshold settings of the analysis. These results indicate that there may be inherent difficulties in reproducing in the lab the more extreme conversational behaviours reported in our everyday life. But with further work to better determine the analysis parameters, this approach can assess how different demands and assistance affects the degree of interaction, an essential facet of group conversation.

Funding: Work was supported by funding from the Medical Research Council [grant number MR/X003620/1].