Polimatrix: Conversation Analysis as a Method for Evaluating Performance in Matrix Games

In this paper we analyse the spoken interactions that take place during matrix games. Matrix games are serious role-playing games, used as training exercises in the military and areas such as strategy studies. Each matrix game simulates a real-world crisis, providing players with specific roles to play and the opportunity to interact in a ‘fail-safe’ environment. While there have been many studies on oral interactions in professional contexts (Kolb 2004, Glenn 2019), there has been less analysis of actual dialogues during table-top gameplay, especially face-to-face negotiation games. In order to address this issue, we designed a game based on the Russia-Ukraine war and recorded sessions of these games in order to compile an annotated corpus. We report findings from the Polimatrix corpus here, looking especially at statistically salient keywords (terminology), n-grams (phraseology) and speech functions (discourse strategies). Generally, we find a correlation between certain linguistic features and contextual variables. For example, the negotiation strategies of (self-declared) ‘novice’ players is typically direct, informal, and preoccupied with the ‘here-and-now’ of gameplay. In contrast, the discourse of ‘expert’ players is typically more respectful of other players’ status, stylistically more elaborate and centres on ‘higher’ strategic themes. Given such differences of linguistic performance, it is worth examining whether language itself should be a factor in the evaluation of matrix game sessions. Previous studies have shown a causal relationship between linguistic proficiency and negotiation outcomes (Taylor et al 2005, Hayakawa et al. 2019). This suggests that the quality of spoken interaction is a key factor in the evaluation of participants and their performances during negotiations. The question then is how to integrate the analysis of language into the design and practice of matrix games. We argue here that facilitators and their students may find it useful to conduct explicit analysis of linguistic performance in the form of (1) introductory conversation analysis and (2) the ‘post-match’ analysis of extracts from previous gameplay.