In order to enable students to directly face empirical data, summarize translation rules and learn translation skills, this paper studies the basis, motivation and methods of applying research dynamics in translation and teaching.Presenting data in class is the main method of dynamically searching corpora, which enables learners to face enough bilingual data that are easy to choose, and makes translation skills and teaching of translation of selected language items relatively focused. In recent years, the emotional analysis text has attracted academic scientists, and the professionals involved in the research, the use of research methods, and the cultural background related to language have become more and more extensive. In this paper, natural language processing is used to analyze emotions contained in translated texts. Natural language processing not only helps to manage the huge ability of data to e ciently translate text, but also helps to extract the hidden emotions in text translation. It only takes half the effort to achieve the multiplier effect. The multi label classi cation in natural language processing can re ect the information contained in emotion. The translated text is more detailed, which is helpful for further research.