Logical fallacies in cross-border news debates: A case study of Bangladesh-India YouTube discourse
Year: 2026
Type: dataset
Abstract: Online user comment sections play a role in shaping public perception in cross-border political debates, yet their rhetorical structure remains underexplored. We introduce BIFaD (Bangladesh–India Fallacy Discourse Dataset), a large-scale corpus of YouTube comments spanning seven years (January 2019 to January 2026), comprising 376,619 comments from 245,497 users across 1,938 Bangladesh-origin news videos about India. Using LLM-based annotation validated against native Bangla-speaking human annotators, we label comments and examine patterns across topics, user profiles, time, and geopolitical events. Fallacious discourse concentrates in geopolitically sensitive topics (International Affairs, Military, Politics, and Crime) and are dominated by Appeal to Emotion, Tu Quoque, and Appeal to Fear fallacies, with sharp event-driven spikes (e.g., 12,177 fallacious comments in the week after the May 2025 India--Pakistan military conflict). While the rate of fallacious comments is weakly related to video-level views and likes, it is positively associated with discussion intensity within comment threads, suggesting fallacy-heavy comment sections generate greater peer interaction. Interrupted time-series analysis across 13 geopolitical events during the period shows that conflict-framed, religiously charged, and diplomatically contentious events produce the strongest increases in fallacy rates. Linguistic analysis using allotaxonometry, ousiometrics word-shift graphs, and Zipf and Heaps scaling shows that fallacious comments are more lexically concentrated and more strongly tied to hostile, identity-based, and conflict-oriented vocabulary, using more dangerous and aggressive language and less good and powerful language than non-fallacious comments. Together, our findings position BIFaD as a resource for studying fallacious reasoning in low-resource, cross-border online political discourse.
Source: Figshare
Language: English
Subfield: Sociology and Political Science
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7204136758