Quantitative Syntactic Features Of Genres From Multi-perspectives

With the substantial development of new technologies in the digital age,quantitative methods based on computational science become increasingly common in genre studies.There has been a surge of interest thereafter in this field.Nevertheless,these studies seem to be insufficient to the best of our knowledge in terms of the following three aspects.Above all,previous research has primarily focused on lexical information and statistical properties of lexical and syntactic features across genres,somehow neglecting characteristics of the internal syntactic relationship.Furthermore,based on previous literature,genre studies regarding three perspectives merits explicit attention.Syntactic features of genres across different languages,of genres across different media within the same language,and of authors within the same genre may all display differences(or not).Last but not least,quantitative linguistics provides dynamic analyses and descriptions on linguistic phenomena,properties of linguistic elements and the relationship among them in order to discover linguistic laws which can be formulated mathematically.Nevertheless,few studies systematically applied methods of quantitative linguistics to genre studies.Therefore,the present study intends to investigate English genres' syntactic properties from a quantitative linguistic view at both the macro and micro level by starting with the cross-language comparisons,and then moving towards cross-and intra-genre(cross-author)ones successively.Three main research questions are as follows: 1)What are the differences and similarities in the syntactic characteristics of different genres across different languages,i.e.,English and Chinese? 2)What is the relationship between syntactic features of different genres within the same language,namely,genres available through the Internet media and traditional media? 3)What are syntactic properties of different authors,Hillary and Trump,across genres? What's more,can syntactic features be a useful classifier in the three aforementioned research questions respectively?Results show that all of the genres show dynamic properties in the cross-language,cross-and intra-genre(cross-author)comparisons.This unveils a finding that genres share both the differences and similarities at the syntactic level,which indicates that language is a self-regulating and self-organizing dynamic system.Major findings are as follows:1)The cross-linguistic comparison shows that,firstly,most of the syntactic features show significant differences across genres between English and Chinese.They can be useful criteria in distinguishing cross-linguistic genres.Secondly,a number of genres in two languages demonstrate similarities at the syntactic level,despite the fact that English and Chinese belong to different language families.Thirdly,language affects genre to some extent.Chinese tends to be governor-final while dependents tend to be on both sides of governors in English.Consequently,the distributions of dependency direction in different genres remain similar within one language.Fourthly,language also plays a role in the extent of difference across genres.Genre variation in English is larger than does in Chinese.This has to do with features of different languages.2)The cross-genre comparison uncovers the following results.Firstly,most of the syntactic variables present significant differences between the web genres and traditional written genres.These features can well distinguish several web genres from traditional written ones.Secondly,being available through different media,web genres differ from traditional written genres to some extent.Thirdly,even the same genre would exhibit differences across different syntactic features.Regarding most of the syntactic variables,such as sentence length,email and twitter are obviously dissimilar from news page;nevertheless,their distributions of certain specific syntactic relations,i.e.,noun compound modifiers,are rather similar to those of news page.Fourthly,most of the web genres share many similarities with traditional written genres.Sentence lengths measured in terms of words in news page and personal blog are close to the end of traditional news in the genre spectrum,and those of email and twitter are similar to the end of traditional fiction;whereas personal blog stands in between.What should be noted is that most of the twitter's syntactic features are clearly distinct from other web genres and traditional written genres.3)The intra-genre or cross-author comparison shows that,firstly,most of selected variables' differences reach statistically significant level,separating some of Hillary's and Trump's genres from the rest of ones.Secondly,genre exerts marked influence on authors' style.Thirdly,the extent of genre effect varies across authors.Some authors are influenced by genre to a larger degree;conversely,others' personal style somehow outshines genre restrictions.Differences of various syntactic properties between Trump's campaign speech and presidential debate are much greater than those of Hillary's.Fourthly,authors' influence cannot be disregarded though.Specifically,Trump employs a heavier usage of numeric modifiers across all genres than does Hillary.In contrast,Hillary tends to use clausal complements and clauses across all of her texts.Their preferences may alter across genres,albeit within a small range.Compared with other authors,their tendency remains similar all the way along.We can see that genres exhibit dissimilarities in terms of syntactic features across different languages,different media,and different authorships.It could not be ignored,however,that no matter what kind of genre,they demonstrate similar regularities with regard to certain linguistic laws.For instance,rank-frequency distributions of sentence lengths across all genres follow the same probability distributions.The distributions of dependency distances for different genres all demonstrate long-tail similarities,signifying a tendency of dependency distance minimization.This may be attributed to universal cognitive mechanism of human languages.The current research examines English genres' syntactic characteristics from multiple perspectives,quantitative results of which may provide objective and statistical support for theoretical opinions,extend the research field of quantitative linguistics and confirm the feasibility of applying syntactic variables to genre studies.During the discussion,the present study made efforts to combine the linguistic features,i.e.,syntactic features,and the possible factors behind them.This indicates that three factors,namely,language,media and author have major influence on genre,which may shed new light on the field of genre studies.Additionally,it may help to further empirical studies in the field of natural language processing and machine learning by implementing parameters of universal linguistic laws.It may also be significant for the practice of automatic genre identification.