Debate and Dialogue: Testing Foundational Assumptions for Studying Linguistic Patterns

Disagreement can take many forms, with outcomes that range from constructive to destructive depending on the approach taken. Among these, debate and dialogue are two approaches that emerge from a distinct set of goals, behaviors, and assumptions. Whereas debate focuses on persuading one another, dialogue focuses on understanding one another. This paper seeks to lay the groundwork for identifying and exploring the linguistic features that differentiate these two approaches, serving as a foundational step toward a larger experiment examining disagreement utterances. The studies test a series of core assumptions critical for designing experiments that can reliably analyze disagreement at a linguistic level. The results demonstrate that speakers can be evaluated both at the utterance level and the speaker level across dimensions identified as debate-like or dialogue-like. Using specifically designed practice examples; the study shows that raters can reliably identify these dimensions through textual data. Next, these dimensions are tested on experimental text data collected in prior studies designed to elicit disagreement. The findings reveal consistent patterns in how raters evaluate speakers and their utterances. Furthermore, the analysis shows that utterance-level data can effectively predict speaker-level ratings within the same conversations, highlighting a significant connection between the two levels of evaluation. By understanding the linguistic features associated with debate and dialogue, this research ultimately aims to provide a framework for individuals to approach disagreements more effectively.