Stanford Film Dialogue Study Reveals Gendered Conversation Architecture

A team led by Samin Khan, Dan Jurafsky, and Shrikanth Narayanan applied network analysis to 4,600 manually annotated directed dialogue events from 38 film screenplays. Their study, titled "You Talkin to Me?", goes beyond traditional word-count metrics to examine who speaks and—crucially—who is spoken to. Using participation shift analysis, chi-squared tests, and paired statistical comparisons across three studies, they mapped the structural flow of conversation. The dataset spans multiple genres, allowing them to isolate gender patterns in both mixed-gender and single-gender scenes.

The results reveal a systemic bias: male characters are disproportionately both speakers and addressees, even in scenes where female characters outnumber them. Cross-gender dialogue was found to be directionally symmetric on average but highly clustered at the film level, indicating that individual movies can skew heavily toward male-dominated exchanges. Interestingly, same-gender turns disperse conversational attention, while cross-gender turns produce tighter dyadic reciprocation. The authors conclude that gender bias in film operates through the architecture of conversation itself—through exclusion from interaction and structural positioning as addressees—rather than through simple imbalances in speaking time. This suggests that even parity in lines does not guarantee equal conversational prominence.