Doughman Jad, Khreich Wael
Suliman S. Olayan School of Business (OSB), American University of Beirut (AUB), Beirut, Lebanon.
PLoS One. 2025 Jan 29;20(1):e0316093. doi: 10.1371/journal.pone.0316093. eCollection 2025.
Historically, the film industry has been male-dominated both in front of and behind the camera, resulting in a longstanding gender imbalance in storytelling and representation. This legacy of male-centric narratives may unconsciously influence critics' expectations and judgments. Existing literature suggests that negative critiques in movie reviews can significantly impact actors' earnings by diminishing a film's commercial prospects. This influence extends to potential reductions in back-end compensation and marketability for future projects, and it can directly affect actors' well-being, leading to increased stress levels and elevated cardiovascular reactivity. Previously, gender biases in movie reviews were computed using disparities in male-led versus female-led movie ratings or box-office earnings; however, no work has been done to quantify the linguistic biases within movie review transcripts. This work aims to leverage our language-model-powered gender bias detection system to measure benevolent sexism, hostile sexism, explicit marking of sex, dehumanization, and generic pronouns in reviews published by professional critics. Therefore, we analyzed 17,165 professionally written reviews, comprising a total of 735,000 sentences. Our analysis uncovers pronounced representation bias in key movie roles, with 72% of first actors, 91% of first directors, and 86% of first writers being male. More importantly, the findings indicate that movies with female-dominated casts evoke, on average, 149% higher magnitude of hostile sexism and 44% higher magnitude of benevolent sexism in their reviews compared to movies with male-dominated casts. We also find that benevolent sexism is most common in Family and Music genres, reflecting the romanticization of gender roles, whereas hostile sexism peaks in Romance. A non-parametric statistical analysis revealed significant gender differences in benevolent and hostile sexism scores for movie reviews, with female first actors, directors, and writers receiving higher levels of both benevolent and hostile sexist criticism compared to their male counterparts.
从历史上看,电影行业在镜头前后都由男性主导,这导致了在故事讲述和呈现方面长期存在性别失衡。这种以男性为中心的叙事传统可能会在不知不觉中影响评论家的期望和判断。现有文献表明,电影评论中的负面批评会通过削弱电影的商业前景,显著影响演员的收入。这种影响还会延伸至后端报酬的潜在减少以及未来项目的市场适销性,并且会直接影响演员的幸福感,导致压力水平上升和心血管反应增强。以前,电影评论中的性别偏见是通过男性主导与女性主导电影的评分或票房收入差异来计算的;然而,尚未开展任何工作来量化电影评论转录本中的语言偏见。这项工作旨在利用我们基于语言模型的性别偏见检测系统,来衡量专业评论家发表的评论中的善意性别歧视、敌意性别歧视、性别明确标注、非人化以及通用代词。因此,我们分析了17165篇专业撰写的评论,共计735000个句子。我们的分析揭示了关键电影角色中存在明显的呈现偏见,72%的主角演员、91%的导演以及86%的编剧为男性。更重要的是,研究结果表明,与男性主导演员阵容的电影相比,女性主导演员阵容的电影在评论中引发的敌意性别歧视平均高出149%,善意性别歧视平均高出44%。我们还发现,善意性别歧视在家庭和音乐类型中最为常见,反映了性别角色的浪漫化,而敌意性别歧视在浪漫爱情类型中达到顶峰。一项非参数统计分析显示,电影评论在善意和敌意性别歧视得分上存在显著的性别差异,女性主角演员、导演和编剧比男性同行受到更多的善意和敌意性别歧视批评。