| The international mainstream media is the carrier of producing,organizing and disseminating information,and to a certain extent,it influences the international communication pattern,international discourse and the construction of national image.At present,the international communication pattern is still "strong in the West and weak in the East",and the Western mainstream media dominates the international agenda setting,and its media reporting discourse becomes an important carrier for understanding the world,enhancing cultural soft power and international communication capacity building.The mediated environment brings convenience to international communication,at the same time makes the whole information operation system more complex,which brings both opportunities and challenges to the analysis of international media discourse.Constructing a combination of qualitative and quantitative discourse analysis methods for international mainstream media coverage can dig valuable news clues from the massive media reports data and strengthen international communication capacity building.Based on this,this paper explores the combined qualitative and quantitative international mainstream media discourse analysis methods on the basis of combing topic analysis and sentiment analysis with the existing media discourse analysis methods,to provide methodological guidance for media discourse analysis.This paper includes the following aspects:(1)In terms of research on mainstream media topic analysis methods:(1)Developing rules for dividing the life cycle of mainstream media information.Based on the life cycle theory,this paper proposes a method of dividing the life cycle of media information by combining the "event domain" and the number of information.(2)Improving TF-IDF algorithm to extract topic feature words.This paper improves the TF-IDF algorithm to reduce the influence of word frequency on the weight,and adds factors such as word location and paragraph span in the selection of text features.(3)Constructing the LDA topic analysis model.Specifying the application scenarios,input data,parameters and training steps for model construction,and combining the TF-IDF weights of feature words to filter the feature words under media topics.(4)Exploring the similarity-based topic association analysis.The study of public opinion topic association relations is an important part of topic evolution analysis.This paper formulates topic association relation rules based on similarity awareness,and classifies the association relations among topics into various types such as generation,extinction,inheritance,splitting and merging.(2)In terms of research on mainstream media sentiment analysis methods:(1)Constructing lexicon.It mainly includes a basic sentiment lexicon,negative vocabulary,adverbs of degree and connectives.(2)Selecting emotional feature.Selecting five types of text features,including sentiment words,negative words,degree adverbs,connectives and special symbols.(3)Determining the sentiment feature weighting algorithm.Considering the influence of positive sentiment words,negative sentiment words,negative words,degree adverbs and connectives on sentiment tendency,the sentiment analysis or sentiment extreme value weighting are performed.(4)Constructing the SVM classification model based on sentiment features.The sentiment features are extracted from the media corpus,and the corresponding sentiment feature weights are calculated,and the model is trained based on the sentiment feature values received by the classifier to build the sentiment classifier.(3)In terms of research on mainstream media critical discourse analysis methods:critical discourse analysis theory has strong disciplinary integration,and this paper takes into account both quantitative statistics and qualitative textual interpretations through the combination of critical discourse analysis methods with metrological statistical methods,corpus techniques and rhetorical methods in linguistics,to dig deeper textual information of international mainstream media and improve the objectivity of the study.With the help of lexical analysis software Ant Conc,this paper constructs a multidisciplinary integrated method for analyzing mainstream media discourse by interpreting semantic prosody in media reports’ discourse through word frequency statistics,keyword analysis,collocation analysis,critical metaphor analysis,combined with the critical discourse analysis method.(4)Empirical study.In order to examine the discourse of international media coverage,the paper selects influential and representative international mainstream media and takes the "COVID-19" as an example.The empirical results founds that:(1)In terms of topic analysis,the evolutionary cycle of media information on the events of the COVID-19 is divided into four periods: incubation period,outbreak period,spreading period and elimination period.Combining the improved TF-IDF algorithm and LDA topic analysis model can better mine the topics of media,and the similarity-based topic association analysis method can fully explore the association relationship of topics in different time slices and identify the evolutionary path of topics.(2)In terms of sentiment analysis,this paper proposes an international mainstream media sentiment classification method based on the combination of sentiment dictionary and support vector machine.Through experiments with multiple feature combinations,the impact of different features on the SVM classification model is evaluated,and the optimal feature combination is "sentiment word + negative word + degree adverbs + associated word + special symbol".(3)In terms of critical discourse analysis,the combination of word frequency statistics and keyword analysis reveals that different media reports have different issues and positions on the same event,and the collocations are mostly verbs,and the collocation structures are mainly subject-predicate and verb-object structures,there are inconsistencies between the collocations and the emotional tendencies of index lines,and the critical metaphor analysis can interpret and show the rhetorical phenomena in the reports.The innovations of this paper are mainly reflected in the following aspects:(1)In terms of research methodology,a combined qualitative and quantitative method for analyzing international media coverage discourse is constructed.This paper applies topic analysis,sentiment analysis and bibliometric methods to the analysis of international media coverage discourse,and constructs a method for analyzing media discourse from an interdisciplinary perspective.Through the combination of quantitative word frequency statistics and qualitative critical discourse analysis,it makes up for the subjectivity deficiency of critical discourse analysis and strengthens the scientific and empirical nature of the method of media coverage discourse analysis.(2)In terms of research perspectives,there are few domestic studies on discourse analysis methods for media coverage,and there is a lack of methods applicable to the identification and analysis of international media discourse.This paper adds factors such as word position,paragraph span,topic,sentiment and discourse to the analysis of text features to effectively explore the discourse concerns in the international media corpus. |
PDF Full Text Request