A Probabilistic Modeling Study of the Dynamics of Discourse Expression and the Construction of Discourse Power in an English News Corpus

On the subject and influence of news is not an unfamiliar topic, however, audiences generally feel that journalists are manipulators of discourse. In this paper, from the perspective of linguistics, we explore the lexical dynamics in different English new corpora and construct a probabilistic model of discourse power for research. Web crawler technology is utilized to obtain relevant corpus information, and the data is processed and handled. Then the Bayesian belief network model is established by determining the common retrocausal reasoning logic of speech act derivation and Bayes’ theorem to link the two, and improved for the condition that their attributes must be independent of each other, and finally the discourse power probability model is constructed based on the Bayesian belief network model. English financial news, the World Customs Organization News Corpus, and the BROWN Corpus are chosen as research objects for empirical analysis. The results of the study show that discourse power construction of conversational meanings can be carried out based on lexical features if it is based on the optimal set of features F1 and F2, and the results of discourse power construction do not differ depending on the contextual discourse or the type of conversational meanings. In terms of lexical usage, the proportion of real words (especially nouns) in financial news is higher than that in generalized discourse, indicating that the information density of financial news is greater than that of generalized discourse, and in terms of discourse function, Customs News English shows a stronger informativeness than generalized and generalized news English.