How Emotions Take Shape: A Multimodal Exploration of Depression on TikTok (Currently in Under Review)

This research examines how depression is represented through multimodal storytelling on TikTok. We collected over 550 videos tagged with #depression and #depressed and applied a computational framework that integrates topic modeling, computer vision, and audio feature analysis. This approach allowed us to trace not only the textual themes of posts, but also the colors, facial expressions, and sonic patterns that together construct narratives of mental health.

Findings reveal eight recurring themes—from heartbreak and trauma to mental health awareness and everyday reflections—and show that creators frequently rely on dark visual palettes, intimate home settings, and trending audio to convey their experiences. While sadness and neutrality dominated facial expressions and tonal analysis, moments of humor, gratitude, and tenderness also surfaced, complicating stereotypes of mental health discourse online. Grounded in theories of representation and identity, this study demonstrates how TikTok enables users to challenge reductive portrayals of depression and instead circulate more layered, affectively rich narratives in the digital public sphere.

• Presented at the Computational Methods division, ICA 2024, Gold Coast, Australia

Stay tuned for publication!