Specific lexical choices in narrative text reflect both the writer's\nattitudes towards people in the narrative and influence the audience's\nreactions. Prior work has examined descriptions of people in English using\ncontextual affective analysis, a natural language processing (NLP) technique\nthat seeks to analyze how people are portrayed along dimensions of power,\nagency, and sentiment. Our work presents an extension of this methodology to\nmultilingual settings, which is enabled by a new corpus that we collect and a\nnew multilingual model. We additionally show how word connotations differ\nacross languages and cultures, highlighting the difficulty of generalizing\nexisting English datasets and methods. We then demonstrate the usefulness of\nour method by analyzing Wikipedia biography pages of members of the LGBT\ncommunity across three languages: English, Russian, and Spanish. Our results\nshow systematic differences in how the LGBT community is portrayed across\nlanguages, surfacing cultural differences in narratives and signs of social\nbiases. Practically, this model can be used to identify Wikipedia articles for\nfurther manual analysis -- articles that might contain content gaps or an\nimbalanced representation of particular social groups.\n