跨文化礼貌研究：基于英语Twitter与中文微博

跨文化礼貌建模有助于通过揭示被认为恰当和礼貌的行为来改善跨文化交流。我们研究了美国英语和汉语普通话中与礼貌相关的语言特征。首先，我们标注了来自美国的5,300条Twitter帖子和来自中国的5,300条新浪微博帖子的礼貌评分。其次，我们开发了一套英语和汉语礼貌特征集「PoliteLex」。结合经过验证的心理语言学词典，我们研究了语言特征与跨文化感知礼貌之间的相关性。我们发现，在中文微博上.Future-focusing conversations, group affiliation identification and gratitude are perceived as more polite compared to English Twitter. Specifically, death-related taboo topics, pronoun absence or poor pronoun choice, and informal language correlate with higher impoliteness on Mandarin Weibo, contrasting with English Twitter. Additionally, we developed machine learning models achieving an F1 score of 0.886 for Mandarin Weibo and 0.774 for English Twitter in politeness prediction.