Abstract: To analyze the semantic sequence in a certain type of discourse, this study proposes a data-driven method of discourse analysis that starts with statistically selected prepositions to extract concordance patterns and semantic sequences.Application of the method to the self-compiled corpus proves that the method is helpful in revealing the macro-level features of the discourse and its diachronic variations.The findings of the present study point to the existence of large quantities of semantic sequences resulting from semantic co-occurrence, which complements phraseological studies that prove the existence of large quantities of phraseology in language.Semantic sequence may be regarded as another important feature of phraseological units.The study of semantic sequence widens the scope of phraseological study and enriches the research on extended units of meaning.