Utterance variation and its stylistic features in Japanese language education using the dominant relevance approach

In this paper, based on the analysis of onomatopoeia in the Japanese language, based on the constructional and grammatical features of the Japanese language, combined with the teaching methods and characteristics of listening, reading and writing in the Japanese language, we utilize the dominant correlation degree method for the characterization of utterance change and its stylistic features. For semantic association, the local dominant association algorithm is employed. The logistic regression algorithm is used to train the two-layer classification model, and logistic regression, plain Bayes and support vector machine are chosen as the analysis and comparison algorithms to judge and analyze the semantics of utterances in Japanese language education. The two-layer classification model was examined to examine the impact of each index, and the word length of written and spoken language was scrutinized for learners and native speakers. The accuracy of each reached more than 99.8%, and the difference was controlled at 0.001. The size order of the word type coverage was Lo>Ns>Lw>Nx>Nr.