What text can tell us about collaborative work: Team shared knowledge and cohesion in an online decision-making task.

  • Latifah Alshammary

Student thesis: Doctoral ThesisPhD


When small teams collaborate on decision-making tasks, their performance depends on how well members share mental models about the task and about the team itself — and on how well they work together, team cohesion. When collaboration happens online, with little extra-verbal communication, individuals’ models must be constructed from teams’ communication texts. This opens up the possibility that the same texts that participants are using could be mined automatically, using computational text analysis techniques.
In this thesis, two studies are reported in which small groups comprising 3 (study 1) or 2 (study 2) individuals collaborated through a bespoke texting interface on the desert survival task. After collaboration, participants answered questions about their task and team models (study 1) or about their group cohesion (study 2). The text collaborations were recorded and analysed using various computational and statistical techniques to test hypotheses concerning the relations between communicative discourse, shared mental models (team and task mental models), team cohesion, and group decision-making performance.
In study 1, we found a relation between shared mental models and team language style. (In particular, positive tone and the use of ‘I’ were associated with more shared models). The findings from study 2 highlighted that the two dimensions of team cohesion: ‘task cohesion’ and ‘team cohesion’ were positively correlated with each other, and both could be predicted by automatic analysis of conversational discourse. (In particular, the use of “we” and the frequency at which task items were mentioned predicted both team and task cohesion.) Furthermore, a direct relation between team performance and some linguistic features were revealed in study1 (but not replicated in study 2). These findings highlight the potential for analysing discourse to predict collaboration and performance, but also the need for more research into the relationship between team language and group dynamics in virtual decision-making teams.
Date of Award16 Jun 2021
Original languageEnglish
Awarding Institution
  • University of Bath
SponsorsRoyal Embassy of Saudi Arabia & University of Hafr Al Batin
SupervisorSimon Jones (Supervisor) & Stephen Payne (Supervisor)

Cite this