Same platform, different stories: TikTok and the battle over immigration narratives

Summary

This paper investigates how immigration is framed on Canadian TikTok and how the platform’s distinctive memetic and interactive affordances shape that discourse. Through a mixed-methods content analysis of 344 high-engagement English-language videos collected in July 2025, the authors find that pro-immigration content outnumbers anti-immigration content roughly three to one — a striking inversion of patterns documented on X/Twitter. They argue that TikTok’s affordances operate as a “double-edged resource”: users across ideological positions deploy the same vernacular features (humor, community Toks, trending audio) in largely symmetrical ways, meaning the platform can simultaneously enable digital inclusion and amplify xenophobia. The dominant axis of contestation in Canada is economic rather than identity-based, reflecting post-pandemic affordability anxieties.

Key Contributions

• One of the first systematic content analyses of immigration framing on TikTok, extending a literature concentrated on X and Facebook.
• A replicable methodology for studying geographically-bounded TikTok content in jurisdictions lacking Research API access, combining the Zeeschuimer scraper with time-varied sampling to mitigate algorithmic personalization.
• Operationalization of the Memetic Interactivity Codebook (MIC) for rule-based analysis of TikTok affordances.
• Empirical bridging of immigration-framing scholarship with platform-affordance theory, showing that identical features support both inclusionary and exclusionary discourse.
• Documentation of generative AI as an emerging vector for racist “speculative worldbuilding” in anti-immigration content.

Methods

The authors scraped 5,305 videos via Zeeschuimer (a Firefox extension) using value-neutral keywords and hashtags (e.g., “Canada immigration”, canadaimmigration), searching twice daily over seven days in July 2025. Purposive filtering — 2025 videos, >100,000 plays, English/unknown metadata, non-US location, most-played video per user — yielded 344 videos after intercoder reliability checks. Stance and frame were manually coded using a 13-frame typology drawn from Helbling, Kelling & Monroe, and Gruzd et al. (Krippendorff’s α > 0.8), while affordances were coded via the MIC. Analysis used chi-square and Fisher-Freeman-Halton exact tests with Monte Carlo simulation and Bonferroni-adjusted post-hoc comparisons.

Findings

• Stance distribution: 41% pro-immigration, 13% anti-immigration, 8% neutral, 38% unrelated — inverting typical X/Twitter patterns.
• Pro-immigration content was dominated by an “other” frame (62%) covering immigration-consultant advice, peer support, and citizenship celebrations; economic benefits came second (14%).
• Anti-immigration content centered on economic costs (47%), with cultural threats (18%) and nationalism (16%) trailing; identity frames together comprised roughly one-third.
• Only one affordance was statistically significantly associated with stance: anti-immigration videos used non-original audio far less (16%) than pro-immigration (44%) or unrelated (57%) videos.
• Duet and Stitch were entirely unused; community Toks (e.g., canadatiktok) appeared similarly (~21%) across stances, often signaling patriotism.
• Humor served as a shared register across ideological camps, supporting both solidarity narratives and racist anti-South-Asian tropes.
• A minority of anti-immigration videos used generative AI to construct dystopian “speculative futures” of Indian demographic takeover.

Connections

This paper sits within a growing strand of TikTok-as-political-discourse research that takes the platform’s content-centric, memetic logic seriously as distinct from network-centric platforms. It connects directly to Gerbaudo2026-fo on TikTok’s role in political communication and to Cabbuag2024-me on the platform’s memetic and vernacular practices — both relevant for understanding how the “shared language” of TikTok affordances can be appropriated across ideological lines.

Podcast

A research-radio episode discusses this paper: Listen