Understanding how social norms vary across cultures can help us build culturally aligned NLP systems. We propose a culture agnostic approach to norm discovery, using moral emotions, shame and pride, to identify examples of normative expectations and extract corresponding social norms. We present the first cross cultural self-conscious emotions dataset, obtained from 5.4K Bollywood and Hollywood movies, along with over 10K extracted social norms. We validate our dataset using native speakers and demonstrate how our dataset reveals variations in social norms that align with the cultural dichotomy observed in these nations e.g., Bollywood movies emphasize shame due to deviation from social roles, and express pride in family honor, while Hollywood shames poverty and incompetence, and takes pride in ethical behavior. Notably, females are shamed more across both cultures and both cultures shame women for violating similar normative expectations.

翻译：理解社会规范在不同文化中的差异有助于构建文化对齐的自然语言处理系统。我们提出一种文化无关的规范发现方法，利用羞耻与自豪等道德情感来识别规范性期望的实例，并提取相应的社会规范。我们构建了首个跨文化自觉情感数据集，该数据集源自5,400部宝莱坞与好莱坞电影，并包含超过10,000条提取的社会规范。我们通过母语者验证了数据集的可靠性，并证明该数据集揭示的社会规范差异与两国观察到的文化二元性相吻合。例如，宝莱坞电影更强调因偏离社会角色而产生的羞耻感，并因家庭荣誉而表达自豪；而好莱坞则羞辱贫困与无能，并以道德行为为荣。值得注意的是，女性在两种文化中均受到更多羞辱，且两种文化均因女性违反类似的规范性期望而对其加以羞辱。