摘要
在认知语言学领域,AI语言模型的出现引发了机器翻译的一场革命。它不仅可以替代专门的翻译软件实现对文本的翻译,而且真正实现了跨语言交流,可用于口语交流、视频会议、社交媒体等多种形式的翻译。AI语言模型通过学习大量的双语文本掌握多种语言的语法、词汇和表达习惯,可以生成更准确、流畅的译文。当前神经网络机器翻译系统可以通过模仿人脑功能,借助计算机模拟海量神经元,从而理解语言并生成高质量的译文。尽管神经网络机器翻译比传统的机器翻译先进,但是还不能避免理解和分析方面的错误,甚至无法完全避免一些常识性的翻译错误。造成这些问题的根本原因在于AI语言模型的预训练忽视了源语言的伦理参数。因此,机器翻译在理解源语言和目标语言时,可以通过设置词语以及词语配对选择的伦理参数,使机器翻译中存在的伦理问题得到有效解决。
In the field of cognitive linguistics,the emergence of AI language model has led to a revolution in machine translation.It can not only replace specialized translation software to translate texts,but also truly achieve cross-language communication of various forms of translation,such as oral communication,video conferencing,social media,and so on.AI language model can generate more correct and fluent translation by learning a large number of bilingual texts,mastering the grammar,vocabulary and expression of multiple languages.The current neural network machine translation system imitates the function of human brain and simulates a large number of neurons with the help of computer so that it can understand the language and generate high quality translation.Although it’s superior to traditional machine translation,it can make some mistakes due to misunderstanding.The fundamental reason for the problem is that the pre-training of AI language model overlooks the ethical parameters of the source language.Therefore,the neural network machine translation can understand the ethical nuance of the source and target languages by setting ethical parameters for word and word pairing selection so as to effectively correct the ethical errors in the translation.
出处
《天津外国语大学学报》
2023年第6期1-10,108,共11页
Journal of Tianjin Foreign Studies University
关键词
机器翻译
AI语言模型
神经网络翻译
伦理参数
machine translation
neural network machine
AI language models
ethical parameters
作者简介
聂珍钊,教授,博士生导师,研究方向:文学伦理学批评、文学跨学科研究
