This study focuses on the textual value of business Japanese literary translations. By employing multimodal analysis, this paper extracts distinctive text features. It incorporates feature layer fusion to seamlessly integrate features from various modalities using an attention mechanism. These integrated features are then processed by a multilayer perceptron, which aids in classifying text-based emotions and extracting relevant emotional data. A specific research case has been selected to explore the communicative and emotional impact of translated texts from the audience’s perspective. The results show that in different modalities, the mean value of the print path is the highest (4.4056), and the movie and television path is the lowest (4.0749) in terms of audience preference. The book’s content is the primary focus of the audience. In terms of emotional value, the audience not only thinks about the obvious thoughts about “life (64)”, “love (143)”, and “sacrifice (74)” in the story but also about “faith” under these thoughts. Faith”, “humanity,” “justice,” “salvation,” and “betrayal,” “betrayal”.