Pragmatic Failures in Machine Translation of Cross-border Comments: An Intercultural Perspective

Abstract

From the perspective of cross-cultural communication, this study analyzes pragmatic failures in machine translation (MT) of cross-border user comments on TikTok and Xiaohongshu. Based on Thomas’s (1983) pragmatic failure theory, we compare original comments with MT outputs (Google Translate, DeepL) and human translations. Pragmalinguistic failures show up as literal or odd renditions of slang, irony, and internet neologisms. Sociopragmatic failures come from MT ignoring cultural taboos, face strategies, politeness norms, and humor—e.g., turning a Chinese self-deprecating expression into an embarrassingly direct English statement, which can cause negative emotions or stereotyping. Such failures hurt genuine cross-cultural understanding and may even trigger unintended offense. The study suggests boosting MT’s cross-cultural adaptability via culture-aware corpora and human-in-the-loop post-editing. This research offers empirical evidence for translation evaluation and cross-cultural communication education in the social media era.