Enhancing discussions through technology requires a deep understanding of their building blocks and dynamics. Disagreement plays a key role in shaping the productivity of discussions. The ways we choose to express disagreement—Disagreement Strategies (DS)—impact both the cognitive-epistemic and social-emotional dimensions of the discussion, hindering or facilitating participants’ learning from each other. Uttering DS inherently fuses these dimensions, resulting in a trade-off where one dimension may diminish, amplify, or overshadow the other. However, we found no theoretical framework addressing the correspondence between DS and discussion productivity in terms of learning. To bridge this gap, we conducted a Systematic Literature Review (SLR) and Qualitative Meta-Analysis (QMA), pursuing two research questions: (I) Which primary forms of DS were reported or can be elicited from the literature? (II) How can they be ranked according to learning? Our findings reveal prominent characteristics of DS studies across a variety of settings, languages, cultures, and populations. We challenge their limited power of focus, often dichotomous, on assessing discourse productivity. Complimented by cluster analysis and statistical testing, the QMA enabled us to create the ‘Hierarchical Taxonomy of Disagreement Strategies’ (HiTODS), comprising 18 DS categorized into 4 unique clusters. The DS are ranked in light of our definition of responsiveness, reflecting the degree to which a speaker is attuned to (fuses) both dimensions. We discuss the implications of the taxonomy as a refined framework for researchers and practitioners. Specifically, by laying the foundations to harness artificial intelligence for automated, real-time analysis of contentious discussions. • A systematic literature review and qualitative meta-analysis on disagreement. • Disagreement plays pivotal, multidirectional roles in discussion productivity. • We conceptualize discussion productivity as temporal, fusing two dimensions. • We created a hierarchical taxonomy of disagreement strategies ranked by learning potential. • Our work sets the stage for AI-based real-time analysis of contentious discussions.
Salman et al. (Thu,) studied this question.