The multimodal discourse analysis method is first analyzed in this paper, and a comprehensive framework is established based on it. Secondly, a diversified English translation instruction model is constructed based on the concept of 'human-centered' education. Finally, we integrated multimodal discourse analysis and diversified English translation instruction and conducted a comparative test, which showed that the mean scores of the experimental class in the post-test were 48.56, 29.36, 9.46 and 9.74 for the total vocabulary and different types of vocabulary use, respectively, while the corresponding figures for the control class were 46.16, 27.84, 9.13 and 9.19. The mean post-test scores of the experimental class were higher than those of the control class. This paper's teaching method can guide students' teaching interaction, enhance students' interest in English translation, and enhance the classroom teaching effect.