A Comparative Study On Character Interaction In Chinese And English Version Of Legends Of The Condor Heroes Based On Social Network

Although much research has been carried out on Digital Humanities(DH)and Corpus Translation Studies(CTS),there is still a critical gap as to how the two research domains integrate and interact with one another.This study attempts to bring the translation studies(TS)from the perspective of DH with the combination of specific methods on both strands,adopting a hybrid design of both quantitative and qualitative approaches and taking as an example of the literary texts of Legends of the Condor Heroes(LCH)(“射雕英雄传”“shèdiāo yīng xióng zhuàn”),a Chinese wuxia classic by Jing Yong,with the aim to better understand the intricate character relationships and their bearing on the different stages of the plot,as well as the influence of translator’s subjectivity on the presentation of the plot.The term DH here refers to text-analysis-related techniques and computational methods,with the former including natural language processing(NLP)and social network analysis(SNA),and the latter the hierarchical cluster analysis(HCA),Wilcoxon Signed-Rank Test and Friedman Test,which are common in the field of DH,data mining and non-parametric statistical test.The research procedures are as follows: firstly,constructing the Corpus of Chinese and English Version of LCH.Word segmentation and POS(part of Speech)annotation are performed on the Chinese text but only the latter on the English with a Python program;Two dictionaries are constructed,a character dictionary and a character alias mapping dictionary for disambiguation of noun and pronoun referents(DDNPR);the data of all characters in the whole book is traversed with a self-developed Python program,the DDNPR is employed to align the names and aliases of the same character into one category with manually checking to ensure the accuracy of character recognition.Secondly,extracting the character interaction(CI)of the whole book and each chapter with a self-developed Python program,employing manual checking to improve the accuracy of the resulted data,and then importing into Gephi to draw the CI graph and calculate relevant network parameters.“CI” refers to the five types of interactions between characters: dialogue,behavior,co-occurrence,mention and emotion.The network parameters refer to five indicators: graphic density,degree,weighted degree,Page Rank,and betweenness.Thirdly,comparing the similarities and differences of book-level CI network in Chinese and English.Fourthly,comparing the similarities and differences of chapter-level CI networks in Chinese and English.Fifthly,discussing the influence of characters on the development of the plot with different influential power(via Page Rank)and bonding power(via betweenness).We first conduct the cluster analysis via Page Rank and betweeness of 40 chapters,and the clusters obtained represent characters with different influential power and bonding power;Borrowing the idea of dividing the “plot” into 5successive stages in Freytag’s Pyramid Structure,and corresponding the stages to the appropriate chapters matching the content of the novel,the 5 stages are ‘Exposition’(Chapter 1-6)-‘Rising Action’(Chapter 7-26)-‘Climax’(Chapter 27-34)-‘Falling Action’(Chapter 35-39)-‘Resolution’(Chapter 40).We calculate the mean value of Page Rank and betweenness of different clusters at 5 stages and compare the differences between Chinese and English;then use Friedman Test to examine whether there were significant differences among them in Chinese and English;if there were differences,we conduct Post-hoc Test to discover the clusters with differences.Four research questions are put forward:(1)What are the similarities and differences of book-level CI networks in Chinese and English?(2)What are the similarities and differences of chapter-level CI networks in Chinese and English?(3)How do characters with different influential power in Chinese and English influence the development of the plot?(4)How do characters with different bonding power in Chinese and English influence the development of the plot?Concerning the book-level CI networks,the graphic density is higher in English than in Chinese,probably due to the deletion of certain character’s scenes in English,leading to the increase of graphic density.The similarities lie in two aspects: both networks consist of6 communities;the degrees and weighted degrees both follow the Power-law distribution,indicating that only a few core characters have traversal interactions with most of other characters,and only a few characters have stronger interactions between themselves.The differences are manifested in the ranking of the 14 typical characters in Chinese and English.For example,the Page Rank of Genghis Khan rises from 10 th in Chinese to 5th in English,while the betweenness of Wanyan Honglie falls from 5th in Chinese to 10 th in English.Concerning the chapter-level CI networks,Wilcoxon Signed-Rank Test find that there is no significant difference in the means of degree,Page Rank and betweeness,indicating that the translators have well preserved the CI interactions,the influential power and bonding power of characters in each chapter;there is significant differences in the means of graph density and weighted degree,with English having a lower graphic density than Chinese but a higher weighted degree than Chinese,indicating that the traversal of CI in English is higher than Chinese,and the strength of CI is also higher than Chinese.The clustering via Pagerank forms 4 clusters.Differences exist in the membership and the member gather-together in each cluster in Chinese and English.The less influential(Core 3)and the least influential(Core 4)cluster in Chinese and English influence in similar manner in the five stages of the development of the plot,and the differences appear in the two clusters with high influence(Core 1 and Core 2).The most influential cluster(Core 1)consists of 7 characters,with “Guo Jing” and “Lotus Huang” taking the lead in Chinese,whose influence fluctuates with the plot and peaks at the “Resolution” stage;In English,the most influential cluster(Core 1)consists of only 2 character of “Guo Jing” and“Lotus Huang”,whose influence peaks at the “Climax” and then falls.The next most influential cluster(Core 2)in Chinese is the 6-person-group of “The Six Freaks of the South”,whose influence shows a U-shaped change and peaks at the “Resolution”;while in English,Core 2 is composed of 5 persons who are highly skilled in martial arts,such as“Apothecary Huang”,“Viper Ouyang”,“Count Seven Hong”,etc.,which peaks at “Rising Action” and then falls,indicating that the translators have strengthened the influence of protagonists to advancing the development of the plot as “Guo Jing” and “Lotus Huang”,and the characters with strong martial arts skills.Friedman Test shows that in Chinese and English there is an interaction effect between characters with different influences and the development of the plot.Post-hoc Test shows that there is significant differences between the two clusters with strong influence(Core 1 and Core 2)and the two ones with weak influence(Core 3 and Core 4)in Chinese;in English,the significant differences exist between the strongest and weakest clusters(Core 1 Vs.Core 4)and the weaker and weakest clusters(Core 3 Vs.Core 4),indicating that the distinction between character’s influence is weaker in English than in Chinese.The clustering via betweenness yield 4 clusters as well.In both Chinese and English,the cluster with strong bonding power are “Guo Jing”(Bond 1)and “Lotus Huang”(Bond2),but difference exist in the membership and members gather-together in the weaker(Bond 3)and weakest(Bond 4)clusters.The two clusters with weak bonding power play similar roles in the five stages,but not the two ones with strong bonding power.In Chinese,both “Guo Jing” and “Lotus Huang” maintain their growth trend until “Climax”,after which “Guo Jing” increases more and peaks at “Resolution”,while “Lotus Huang”decreases significantly;In English,both of them show fluctuating growth,with the bonding power of “Guo Jing” peaks at “Falling Action” and then decreases significantly,while “Lotus Huang” peaks at “Resolution” and surpasses “Guo Jing”.These differences indicate that the bonding power of “Guo Jing” is reinforced in Chinese and “Lotus Huang”in English.Friedman Test shows that in Chinese and English there is an interaction effect between characters with different bonding power and the development of the plot.Post-hoc Test shows that in both Chinese and English there is significant differences between the cluster with the most bonding power and the weak and the weakest clusters(Bond 1 Vs.Bond 3 and Bond 1 Vs.Bond 4),indicating that the differentiation of different character’s bonding power parallel in Chinese and English.The changes in the character’s influential and bonding power in English can be attributed to the translator’s subjectivity,to be specific,the flexible translation strategy adopted in order to conform to the reading habits of Western readers and to keep moving closer to the target language.For example,in the case of the two couples with less influential and bonding power,the translators choose omission to tone down the love line of Cyclone Mei and Hurricane Chen,addition to highlight the love line of Yang Kang and Mercy Mu,and variation(i.e.,adjusting the order of sentences)to delineate in a clearer way the plot storyline of the novel,such as the martial arts competition in Mount Hua between Count Seven Hong,Apothecary Huang and Guo Jing.These exemplify the principles of explicitation and normalization proposed by Baker in CTS.This study explores the translation phenomenon from a DH perspective,combining both “distant reading” and “close reading”,and uses SNA techniques and related computational methods to reveal the influence of CI on the development of the plot,as well as the influence of author’s and translator’s fingerprints on the presentation of plot.It is hoped that this study may trigger more interesting explorations of the integration of DH and CTS.