Podcasts have emerged as a significant medium for political communication, particularly as "low-attention media" that blend affective intimacy with multitasking listening practices. This study analyzes how Taiwanese legislators used podcasts during the 2024 controversy over parliamentary reform and legislative power expansion. Drawing on 28 podcast episodes from Huang Kuo-chang and Miao Poya, the research employs a multimethod design that combines Mandarin-language corpus analysis (Corpro) with reflexive use of generative AI (ChatGPT-4o). Findings reveal contrasting rhetorical styles. Huang's discourse emphasizes emotional immediacy, employing colloquial registers, confrontational humor, and moral polarization to mobilize listeners. Miao's discourse exemplifies institutional rhetoric, privileging legalistic terminology, procedural reasoning, and appeals to constitutional norms. Yet the distinction is not absolute: Huang occasionally invoked procedural arguments, while Miao sometimes adopted affective storytelling. These overlaps demonstrate how political actors navigate a rhetorical spectrum rather than a strict binary. Empirically, the study contributes one of the first systematic analyses of podcast-based political discourse in Taiwan, addressing gaps in non-Western digital media research. Methodologically, it illustrates both the potential and the limits of integrating corpusbased tools with generative AI as heuristic supplements to human-led interpretation. The findings underscore how podcasts' affordances reshape the balance between emotional resonance and institutional authority in digital political communication.
