Sentiment and collaboration in online learning: positive interactions predict higher relational satisfaction

In the evolving landscape of educational and professional settings, effective collaboration emerges as a cornerstone of success, particularly in hybrid environments that blend virtual and in-person interactions. However, the competencies that need to be developed to support effective online collaborations remain elusive. This study employs a 15-category custom sentiment lexicon analysis to distil whether specific interpersonal competencies in pairs of undergraduate students during a naturalistic online seminar could influence both group relational satisfaction and task performance. 30 pairs of undergraduate participants (N = 60) participated in a structured 40-minutes online seminar that included a pre-recorded lecture, a discussion activity and a presentation of the discussed content. To capture the different types of collaborations that could occur in the real-world, three types of between-subject discussion activities were conducted: Cognitive Conflict, Brainstorming and Problem-solving, with 10 pairs assigned to each task condition. Recorded transcripts during the discussion activity were analysed for positive, negative and neutral clusters of sentiments. Presentation slides were scored based on a standardised rubric. Our results showed that the Cognitive Conflict task evoked greater negative sentiments compared to the other tasks. Findings demonstrated that positive sentiments (e.g. providing verbal agreement on task process, words of encouragement, greetings and praise for work completed) were found to significantly enhance group relational satisfaction. These findings have direct pedagogical application, as educators can hone these distilled behavioural competencies to better support collaboration in online learning settings.