A Quantitative Analysis Of Business Narratives Based On The Crtical Genre Analysis Theory

Business narratives refer to non-financial information in business reports.The recent narrative shift in business English studies has provided increasing empirical experiences for quantifying business narratives and their economic value.Nevertheless,previous genre analysis of business narratives may not always be reliable due to vague concepts,insufficient theoretical basis,and unsystematic generic features.Accordingly,this dissertation,based on the critical genre analysis theory(CGA),proposed a quantitative framework to study the Management Discussion and Analysis Report(MD?A),a representative genre of business narratives,aiming to identify its generic features and demystify its mapping relation to corporate performance.The quantitative framework interprets the four discourse levels in CGA,namely,the text,genre,professional practice,and professional culture levels,from the perspectives of intra-lingual,extra-lingual,and combined dimensions,and in this way,constructs an analytical schema that consists of textual,contextual,and combinational analytical paths.In terms of the textual analytical path,it observes and measures lexico-grammatical features highlighted at the four discourse levels,including lexico-grammatical features representative of MD?As,generic structures,identity markers,and corporate cultures.The contextual analytical path studies the intertextuality salient at the four discourse levels,including the manifest intertextuality,and the interdiscursivity of genres,identities,and corporate cultures.The combined analytical path investigates the combinational features produced by the unity of lexico-grammatical features and intertextuality at the four discourse levels.The main corpus for study is the MD?A corpus,with a scale of approximately two million tokens compiled from 118 MD?A texts published in the 2017 fiscal year by 118 domestic U.S.listed companies.The first research objective is to reveal MD?A generic features.For achieving this purpose,the textual analytical path focused on the analysis of lexico-grammatical features at the four discourse levels.It used the Mann-Whitney U non-parametric test,a mixture of the generic structure analysis method and the ordinal Logistic regression model,the factor analysis method,and the TF-IDF feature quantification method,respectively,for features recognition at each discourse level;on this basis,it applied the probability algorithm of product accumulation event to obtain the four lexico-grammatical indices.As for the contextual analytical path,it focused on the analysis of intertextuality at the four discourse levels.It mainly applied the corpus-based top-down generic structure analysis method,ethnographic method,and Dirichlet distribution method for features recognition at each discourse level;on this basis,it used the Cosine and reverse dispersion algorithms to retrieve the four intertextual indices.In terms of the combined analytical path,it focused on the analysis of combined features at the four discourse levels using the probability algorithm of sum event,and obtained the four combinational indices.The second research objective is to identify the mapping relationship between generic features and corporate performance.In order to achieve this goal,the study used the C5.0 decision tree model to identify the mapping generic features on the three analytical paths,and compared their accuracy and reliability.Main findings concerning MD?As' generic features are as follows:First of all,the four lexico-grammatical indices on the textual analytical path indicate that,the prominent lexico-grammatical features at the text level are nominalizations accompanying prepositional phrases,adjectives,and adverbs,which function to expand or compress information;the main lexico-grammatical features at the genre level are prepositional phrases,adjectives and adverbs,which compress or expand information,and pronouns,pro-verbs,modal verbs,parallel forms and negative forms,which function to characterize and evaluate information;the main lexico-grammatical features at the professional practice and professional cultural levels are nominalizations,which fulfill the function of compressing or expanding information.Accordingly,the most conspicuous lexico-grammatical features across the four levels of MD?As are nominalizations which exercise to compress and expand business information.Secondly,the four intertextual indices on the contextual analytical path indicate that,the main forms of manifest intertextuality are the recurrence of keywords and the content reference to other MD?As;the main forms of genre interdiscursivity are the embedding,mixing,and bending of moves drawn from three types of discourse,namely,letter to shareholders,corporate financial analysis report,and legal statement;the main forms of identity interdiscursivity are the embedding and mixing of three professional identities,i.e.,public relations specialists in public relations practice,auditors in regulatory practice,and analysts in financial practice;and the main forms of culture interdiscursivity are the embedding and mixing of three corporate cultures,i.e.,knowledge-,resources-,and human-oriented cultures.Finally,the four combinational indices on the combined analytical path indicate that,lexico-grammars and intertextuality can be used together as the basis for measuring the formation of each discourse level,and both of them are the main components of discourse levels-they synthesize with each other and inosculate as a whole.In terms of the mapping relationship between MD?As' generic features and corporate business performance,we found that on the textual analytical path,only the lexico-grammatical features representative of MD?As and corporate cultures map onto business performance;all the four intertextual features except the identity interdiscursivity on the contextual analytical path mirror business performance;and all the four combinational features on the combined analytical path reflect business performance.Despite the mapping relationship,there is a variation of mapping accuracy and reliability between generic features.In general,generic features on the combined analytical path show the highest accuracy and strongest reliability,followed sequentially by those on the contextual and textual analytical paths.This finding highlights the importance of discourse integrity in understanding discourse functions: Only by combining text-internal resources such as lexical grammars with text-external resources such as intertextuality can we successfully reveal discourse functions emerged from real life.Based on the research findings,we discussed the generic features of MD?As from multi-disciplinary perspectives involving cognitive psychology,discourse analysis,and mathematical statistics,and came to a prototype of MD?As,their narrative attributes,and statistical radar diagram.With a reflection on the mixed methods adopted in the current study,we discussed the necessity and efficiency of applying the probabilitybased mathematical exponential thinking,the meaning-based man-machine collaborative analysis method,and the data-driven analysis method in the course of language context analysis.Besides,we discussed the mechanism by which economic value emerges from the discourse from two aspects,i.e.,the channel to transmit economic signals and the theoretical basis.This research contributes to the literature from theoretical,methodological,and practical perspectives.Theoretically,it laid a potential theoretical basis for quantifying business narratives and their generic features.In the methodological aspect,it structured a quantitative framework for measuring business narratives,improved methods for quantifying linguistic concepts such as manifest intertextuality and interdiscursivity,created multiple corpora,and developed R scripts for quantitative genre analysis;in this way,it extended the research into the mapping relationship between discourse and corporate performance.In terms of practical contributions,it identified the parameters of MD?As' prototype which are key to teaching innovation under the trend of digital humanities.It also built classification tree models with solid theoretical basis,which would inspire intelligent language analysis in business practice.Nevertheless,the current research has some limitations in its quantitative framework and analysis of the cultural wordlist and lexico-grammatical features.We expect that future study would investigate genetic features not mentioned presently,extend business narratives to the Chinese context,and on this basis,frame language polices that help Chinese enterprise take their place on the international market.