语境与行为意义观视域下机器翻译的边界与困境

The Boundaries and Predicaments of Machine Translation: A Contextual and Behavioral Meaning Perspective

摘要: 机器翻译在技术层面取得了长足进步，但在处理富含语境敏感性、动态性与具身行为性的文本时，仍面临语义表征不充分、语用失效等深层困境。从后期维特根斯坦的“语言游戏说”、奥斯汀的言语行为意义理论和格赖斯的会话含义理论出发，系统审视机器翻译的能力边界与本质局限，可以发现，机器翻译目前面临的困境并非单纯的技术问题，而是根源于人工智能语言的形式化本质与自然语言语义的语境性、动态性、具身行为性及心理意向性之间的根本性差异。

Abstract: Machine translation has made significant technological advances, yet it still faces deep-rooted predicaments such as insufficient semantic representation and pragmatic failure when processing texts rich in contextual sensitivity, dynamism, and embodied-behavioral characteristics. Drawing upon Wittgenstein’s late philosophy of “language games”, Austin’s speech act theory, and Grice’s conversational implicature theory, this paper systematically examines the capability boundaries and essential limitations of machine translation. The findings reveal that the current predicaments faced by machine translation are not merely technical issues, but are originating from a fundamental semantic divergence between the formalized artificial intelligence language and contextual, dynamic, embodied-behavioral, and psycho-intentional based natural language.