Materials selection

We choose the Mencius translated by James Legge (1861), Leonard A.Lyall (1932) and D.C.Lau (1970) as the corpus, and the collection of these translations follows the principles of representativeness and comparability. First, the three translators are from different historical periods and have different cultural identities, determining their respective translation motives, translation strategies and distinctive translations. Second, the three translations were translated from the same original text-the annotated version of Mencius by Zhuxi, indicating that these translations are highly comparable. Finally, prior to the corpus analysis, the author excludes the introduction, footnotes and other explanatory materials to ensure the thoroughness of the analysis.

Research tools

This study employed corpus tools such as Tokenizer, Tree Tagger, WordSmith 8.0, AntConc and Readability Analyzer. Tokenizer is a software aimed at word segmentation. Tree Tagger using the Pennsylvania Tree Library code set is an automatic part of speech tagging software developed by Helmut Schmid at the University of Stuttgart, and it can automatically identify and label the part of speech of each word in English, German, French and Italian languages. WordSmith 8.0 is an integrated suite of programs developed by Mike Scott of Oxford University Press, including Wordlist, Keyword and Concord programs, which can produce a list of all the words, phrase or word clusters and examine how words behave in texts. Once the translated texts have been processed using WordSmith 8.0, the lexical features and syntactic features will be revealed. Moreover, AntConc is a tool that can be employed to search all kinds of words in the corpus, and Readability Analyzer can be utilized to count the passive sentence information and readability information of texts.

Measurement

The specific linguistic features of the translated text mainly include lexical, syntactic and textual features [28].

The lexical features include the number of tokens, types, type/token ratio (TTR), standardized TTR (STTR), word length distribution, mean word length (MWL) and lexical density. The type/token ratio (TTR) reflects the richness of text vocabulary. Generally speaking, the higher the TTR is, the higher the vocabulary richness will be. Nevertheless, when the text length is long, the standardized type/token ratio (STTR) that refers to the type/token ratio every 1000 words is usually employed to compare the vocabulary richness among translated texts.

The word length distribution is the probability of the distribution of the word length (i.e., the number of letters that make up the word) in the text, and MWL refers to the average length of words, indicating the lexical complexity to some extent. WordSmith 8.0 can count the number of words of each length in a text and calculate the average word length. LD is the proportion of meaningful words in the entire text of a translation, which can reflect the amount of information contained in the text. It is an important indicator for analyzing the style of the translation. The higher the lexical density is, the greater the proportion of content words in the translation is, the greater the information carrying capacity is, and the relatively more accurate and complete the expression of the original meaning is.By contrast, the lower the lexical density is, the more functional words there may be in the translation, and the lower the informative ability of the text is. The content words in English usually include verbs, nouns, adjectives, adverbs, and numerals. It can be calculated by the following formula. The syntactic features include the number of sentences, average sentence length (ASL) and standard derivation of word numbers between sentences (STD). ASL and STD can reflect to some extent the complexity of syntactic structures, and are often considered as important makers to measure the language characteristics of the translated text and the translator’s style. Generally, the shorter the STD is, the simpler the syntactic structure of the text is, while the longer the STD is, the more complex the syntactic structure of the text is. The STD can be calculated by the following formula. The textual features are the overall expression of the translator’s style, which can be described from numerous aspects. The most important indicators are cohesion and readability information. Cohesion is a fundamental attribute and important feature of one discourse, reflecting the overall and hierarchical sense of the discourse structure. As an important cohesive device, conjunctions enable various logical relationships between clauses in the discourse, helping readers analyze the semantic relationships between sentences and more effectively understand the text. In this study, we manually annotate various conjunctions in the text, use AntConc software to search them individually in each translation, summarize different types of conjunctions, and subsequently calculate their numbers. Text readability refers to the difficulty of understanding a text, which can be employed to test whether the translation is easy for readers to understand at a macro level. The Readability Analyzer can detect the difficulty of reading a text by calculating the Flesch Readability Index of one text. The calculation formula of Flesch Reading Index is as follows: (ASW refers to the average monosyllable numbers of words. The higher the Flesch Reading Index is, the less difficult the text is.)

Procedure

After determining the English texts and corpus software needed for the research, we began to clean the corpus, segment the text words and tag the parts of speech of texts.

Word segmentation and corpus tagging.

First, we import the processed corpus into the Tokenizer application, click the convert button to obtain the segmented file and save the files as Jame Legg’s Mencius 2.txt, D.C.Lau’s Mencius 2.txt, and Leonard’s Mencius 2.txt respectively.

Second, we open the WordSmith8.0 software; enter the WordList interface; click the File button; select Jame Legge’s Mencius 2.txt, D.C.Lau’s Mencius 2.txt, and Leonard’s Mencius 2.txt files; and click “make a word list” button to enter the statistics interface. Thus, we can obtain the vocabulary and sentence information of the three translations as Table 1, 2 and 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Lexical statistics features of the three translations of Mencius.

https://doi.org/10.1371/journal.pone.0305894.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Main distribution statistics of word length in the three translated versions of Mencius.

https://doi.org/10.1371/journal.pone.0305894.t002

Third, we open the Tree Tagger for Windows 3.0 software; click the file; open the File button in the upper left corner; select the three files; set the language to English, the encoding to ANSI, and output form to Word_POS; and then click the Run Tagger button to get the TAGGER files. Subsequently, we name the files James Legge’s Mencius TAGGER.txt, D.C.Lau’s Mencius TAGGER.txt, and Leonard’s Mencius TAGGER.txt.

Fourth, based on the part of the speech coding table of Tree Tagger, AntConc’s Concordance retrieval function is used to search the number of content words, as shown in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Tagging statistics and LD of the three translations of Mencius.

https://doi.org/10.1371/journal.pone.0305894.t003

Fifth, we manually annotate and count the conjunctions in the documents, as shown in Table 5.

Sixth, we use the Readability Analyzer software to obtain readability information about texts, as shown in Table 6.

Lexical features

Tables 1 and 2 display the lexical statistics produced by WordSmith Wordlist.

Based on the data results, D.C.Lau’s translation of Mencius contains the most tokens and types, followed by Legge and Leonard’s translations. All three translations are longer than the original text of Mencius containing 36,970 tokens, which indicates the three translations tend to highlight the implicit information present in the ancient text, making readers comprehend the text more easily. Regarding the STTR analysis, it was found that the STTR of Legge’s translation is 37.92, higher than 35.83 that of D.C.Lau’s translation, showing that Legge adopts a more diverse lexicon in his translation. The STTR of Leonard’s translation is 34.53, showing Leonard uses the narrowest vocabulary. There is a longer MWL and a higher percentage of words longer than eight letters than those of Leonard’s version in D.C.Lau’s translation, suggesting that the words are longer and relatively more complex in D.C.Lau’s translation.

The author calculates the number of content words in the three translated texts of Mencius, and Table 3 presents the results.

The LD of the D.C.Lau, Legge and Leonard’s translations are 59%, 56% and 53% respectively, all of which are over 50%. Among them, the LD of D.C.Lau’s translation is the highest, followed by that of Legge and Leonard’s translation. The result of Laviosa’s research on the LD of the English source language corpus and the target language corpus shows that the LD of the former is 54.95% and the latter is 52.87% [28]. The LD of the three translations is higher than that of the target language, especially that of D.C.Lau’s translation far exceeds that of the source language. Thus, it can be concluded that D.C.Lau’s translation has the largest amount of information compared with the other two translations. There is the lowest lexical density in Leonard’s translation, which indicates that Leonard chooses to add functional words to simplify the translation and make the translation structure more clear and easy to understand.

Syntactic features

The WordSmith Wordlist Program is also used to calculate the number of sentences, average word numbers per sentence and standard deviation of word numbers between sentences, revealing the syntactic features of each translation. The statistics are shown in Table 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Data features of three translations at sentence level.

https://doi.org/10.1371/journal.pone.0305894.t004

According to the statistical data of WordSmith 8.0, the result shows that the average sentence length of translations by James Legge, D.C.Lau and Leonard are 18.42, 17.32 and 17.26 respectively, indicating that the average sentence length of Legge’s translation is the longest, and that of Leonard’s translation is the shortest. From the perspective of language, there are many short sentences in the original text of Mencius and the style is concise. Leonard uses short sentences to produce his translation, which reflects the characteristics of the source language. The average sentence length of Legge and D.C.Lau’s translations are relatively longer, probably because the two translators emphasize the expression of information and transform the conversational language of the original text into narrative language. The STD of Legge’s translation is 13.71, higher than that of the other two translations, which indicates that the sentence length structure is more diversified.

Textual features

We have conducted a statistical analysis of the usage of conjunctions in each translation and compared the overall readability among three translations. The results are shown in Table 5 and 6.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. The use of conjunctions in three translations.

https://doi.org/10.1371/journal.pone.0305894.t005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Overall readability of three translations.

https://doi.org/10.1371/journal.pone.0305894.t006

Table 5 reveals that there are certain similarities and differences in the use of conjunctions in the three translations. First, both translators can clearly convey the various semantic relationships implied in the original text to the reader by adding various conjunctions (e.g.,coordinate conjunction, causal conjunction, temporal conjunction, and conditional conjunction) in the translation, which is consistent with the finding that the tokens of three translations are greater than that of the original text. However, we can also clearly see that D.C.Lau used a higher percentage of conjunctions in translation compared to Leonard and Legge, indicating that he tended to simplify linguistic differences between the original text and the translated text to make the translation easier for readers to understand. It is the use of more conjunctions that results in the longer average sentence length of Lau’s translation.

Simultaneously, we can find that the readability of Leonard, D.C.Lau and Legge’s translations are 79.81, 76.13 and 73.81 respectively, indicating that Legge’s translation is the most difficult to understand and this may be related to the diversified vocabulary, complex sentence structure and fewest conjunctions in Legge’s translation. The readability of Leonard’s translation is the highest, indicating Leonard’s translation is the easiest to read, which is consistent with his translation’s characteristics of the simplest words and sentence structure and the most conjunctions.

In summary, we can see that Legge’s translation has the highest STTR, longest MWL and highest lexical density, indicating the vocabulary is richer and more diverse and his version has the longest mean sentence length with the highest standard deviation, pointing to a complex syntactic structure. Conversely, he uses the fewest conjunctions, making his translation relatively challenging to read. Leonard’s translation has the least varied vocabulary and the shortest MWL with the shortest mean sentence length, making the sentence structure simple. D.C.Lau’s translation has the highest lexical density, indicating a sizable information load and a higher mean sentence length with a greater standard deviation, suggesting a relatively complicated and variable syntactic structure. Meanwhile, the higher percentage of conjunctions and the highest readability of Leonard’s translation indicates that the text is easiest to understand. We can see the following example obtained from the translations.

ST1: 若杀其父兄, 系累其子弟, 毁其宗庙, 迁其重器, 如之何其可也?

[29]

TT1: But you have slain their fathers and elder brothers, and put their sons and younger brothers in confinement. You have pulled down the ancestral temple of the State, and are removing to Ch’i its precious vessels.

[30]

TT2: Can it be right to kill their fathers and elder brothers, bind their young men, pull down their ancestral temples, and carry off their heavy ware?

[31]

TT3: How can it be right for you to kill the old and bind the young, destroy the ancestral temples and appropriate the valuable vessels.

[32]

The use of separate verbs in English is usually on a more formal occasion.There are many dialogs between Mencius and various princes in Mencius, which is a type of dialog genre with a kind of relatively colloquial style. When translating such verbs, using phrasal verbs is reasonable. Respecting the colloquial language characteristics of the original text, James Legge uses two separate verbs (slain and removing) and two phrasal verbs (put… in confinement and pull down) to explain the original text, suggesting the richness and diversity of vocabulary. The sentence structure of Legge’s translation comprises three compound sentences, while there is only one interrogative sentence in D.C.Lau and Leonard’s translations, indicating the complexity of sentence structure in Legge’s translation. D.C.Lau uses a series of single verbs “kill, bind, destroy and appropriate” to translate the single verb in the original text, turning the positive sentence pattern of the original text into a special question guided by the word “how”. It accurately conveys the meaning of the original text and enhances the translator’s condemnation to those immoral behaviors. Leonard, conversely, uses simple and phrasal verbs to convey the meaning of the original text, making the sentence structure simplest and the translation easiest to read.

The missionary in the 19th century

In the early 19th century, a large number of missionaries entered China to study Chinese culture in order to smoothly preach. James Legge (1815–1897) was a missionary of the London Missionary Society in the 19th century. In 1838, he entered the University of London to study Chinese and arrived in Hong Kong, where he served as director of Anglo-Chinese College and devoted himself to spreading Christianity and learning Chinese. After constant contact with the Chinese people and extensive reading of Chinese classics, Legge was deeply impressed by the profound ancient Chinese culture and began to translate The Chinese Classics, which was successfully published in 1861 [33]. In the preface, he wrote the following:

I look at Chinese philosophic classics and am eager to learn about its language, history, literature, ethics and social formation. It seems to me that a Westerner, especially a missionary, can not be qualified to assume the responsibilities required by his present position as a missionary until he has mastered the Chinese classics and examined all the fields of thought of the ancient Chinese sages [30].

The purpose of Legge’s translation work is to open the door to Chinese culture for missionaries and serve missionary work based on deepening the understanding of traditional Chinese culture. In 1893, in the first volume of the revised edition of The Chinese Classics, we find Legge’s high praise for Confucius and Mencius, he once said the following:

I hope I have not treated them unfairly, and the more I study their characters and opinions, the more I respect them. Their influence on the Chinese people is great and beneficial, and their teachings are also important lessons for those of us who profess to be Christian so that faithfulness to the original Chinese classics is the first principle. Not that I am indifferent to the value of an elegant and idiomatic rendering in the language of the translation, and I hope that I was able to combine in a considerable degree correctness of interpretation and acceptableness of style [34].

Therefore, “faithfulness” is the first principle of Legge’s translation of the classics. He used the foreignization translation strategy and the literal translation method to complete his translation. He not only pursued the formal consistency between the translation and the original text but also used extensive and detailed annotations as well as indexes to further express his understanding of the work. In his translation, he deliberately broke target conventions by retaining something of foreignness of the original, hence, the translation has the characteristics of diversified vocabularies, complex sentence structures and low readability.

We can see the following example extracted from the Mencius translation by Legge:

ST2: 孟子曰:地方百里而可以王

[29]

TT4: Mencius replied, ‘With a territory which is only a hundred li square, it is possible to attain to the royal dignity.’

[30]

TT5: Mencius said, A land of a hundred square miles may rule the kingdom.

[31]

TT6: ‘A territory of a hundred li square,’ answered Mencius, is sufficient to enable its ruler to become a true King.

[32]

From the perspective of sentence pattern, Legge completed the translation with two clauses, and the sentence structure is the longest. Translators have different translations of “里(li)” that is a unique Chinese unit of measurement. Legge and D.C.Lau translated “里” as “li”, and they explained the word “li” in their notes, accurately conveying the connotation of the Chinese unit. Leonard chose to translate “里” using the word “mile”, which is familiar to foreigners. The advantage of this translation is that it is convenient for foreigners to understand; However, it is easy to westernize the unique Chinese cultural words, which causes unnecessary association, and is not conducive to the spread of traditional Chinese culture.

James Legge devoted his whole life to the translation and interpretation of Chinese classics, realized the transformation from a missionary to a sinologist and opened up the history of the spread and acceptance of traditional Chinese Confucianism in the West with Chinese classics as the carrier. Therefore, his translation is widely regarded as the authoritative translation.

The customs officer in the first half of the 20th century

Zhan [35] pointed out: “From the middle of the 19th century to the first half of the 20th century, foreigners working in the Chinese Customs gradually realized that despite of different cultural backgrounds, they must be familiar with the Chinese language, which was an important step for the customs to gain a firm foothold and expand its foundation, so they consciously or unconsciously acted as a bridge to introduce the East to the West and opened the curtain of ancient China for the Western world. It makes the Western world feel the charm of ancient Chinese civilization, and promotes the development of international Sinology research.”

Leonard A.Lyall (1867–1934) is a customs officer who has worked in the Chinese customs for 40 years and made various achievements in translating ancient Chinese classics. In his spare time, he translated the Confucian classics The Analects of Confucius (1909), The Mean (1927), and Mencius (1932) into English, and assisted the German Karl E. G. Hemeling in compiling the English-Chinese Dictionary to facilitate foreigners to get familiar with and understand the traditional Chinese characters [36]. In the preface to the English translation of Mencius, Leonard A.Lyall said the following:

I have tried to make my translation as nearly as possible a word-for -word rending of the Chinese to reappear the simple style of Mencius. Where I have failed, I have usually given the literal translation in the notes.

The most important Chinese words have always been translated by the same English words;e.g. jen, love; chun-tzu, gentleman; tao, the Way; yi, right, or once or twice justice. The only exception is li, which I have sometimes translated in good form, sometimes courtesy, and sometimes ceremony [31].

Leonard’s translation can be observed to be primarily literal, emphasizing the consistency of word meaning and sentence structure with the original text and striving to present the concise style of the translation with simple and vivid words and short colloquial sentences; therefore, the translation has a strong readability. However, he overemphasized the literal meaning of words and did not accurately convey the connotation of the original text to some extent. In order to reduce the cultural strangeness of Western readers and draw them closer to the original work, Leonard adopted the domestication translation strategy. That is, he attempted to find similar words familiar to Westerners in English to translate unique cultural words, ignoring the unique meanings of words in specific contexts and reducing the fidelity of the translation to the original work.

ST3:《诗》云“昼尔于茅, 宵尔索綯; 亟其乘屋, 其始播百谷”

[29]

TT7: In the day-light go and gather the grass,

And at night twist your ropes;

Then get up quickly on the roofs;

Soon must we begin sowing again the grain.

[30]

TT8: Out in the reeds by day;

Binding, plaiting by night;

Climb the roof betimes;

There are hundreds of seeds yet to sow.

[31]

TT9: In the day time they go for grass;

At night they make it into ropes;

They hasten to repair the roof;

Then they begin sowing the crops; [32]

Leonard used simple words and short sentence patterns to explain the original text and retained the concise style of the original text. The biggest difference among three translations is that the translation of “亟其乘屋 (ji qi cheng wu),” both Legge and Leonard took the original meaning of “乘(cheng)” and translated it as “quickly climbing on the roof.” Although the translation can reflect the literal meaning of the original, it broke the relationship between the verses and was not easy for readers to understand. However, D.C.Lau took the extended meaning of “乘(cheng)” and translated it into “hurry to repair the roof” which is more accurate and understandable.

In contrast to Legge’s extensive footnotes, there are only a few brief notes in Leonard’s translation. In the preface of The Analects of Confucius, he explained the following:

Since I am addressing the general English audience, too long an extraneous note will only look out of place [31].

Although there are few annotations, the most detailed and clear catalog is shown in Leonard’s translation. This translation includes one table of contents, readers can easily find their own interest in the content pages just by flipping it. Leonard A.Lyall always puts the acceptance of ordinary readers in the first place, therefore, his translation is more popular and widely welcomed by ordinary readers.

The devoted scholar in the second half of the 20th century

In the middle and late 20th centuries with the transfer of Western sinology centers to the United States, American increased its investment in Sinology research, and the English translations of Chinese philosophical classics published have a wide range of coverage and influence. During this period, Chinese scholars in the United States became an important force in the English translation of Chinese philosophical classics, because Chinese translators had a more accurate and profound understanding of Chinese classics and were able to gradually present the philosophical connotation and cultural essence of Chinese classics to American readers [37].

D.C.Lau (Liu Dianjue,1921–2010) is a famous sinologist, translator and philosopher. In 1950, he graduated from the Glasgow University and taught at the London University. During his 30 years of teaching, he has devoted himself to the study of Chinese classics as well as translation work. In the 1960s and 1970s, D.C.Lau translated Tao Te Ching (1963), Mencius (1970) and The Analects of Confucius (1979) for the Penguin Classics series. He once recalled the origin of translation:

After I twice refused with the publication of other people’s translations of Tao Te Ching, the editor was a little unconvinced, and he wrote to me and said, “Since you said that they can’t translate it, why don’t you try it yourself?” This is why I translated Tao Te Ching, At that time, the publishing industry was flourishing, they wanted to develop the section of Chinese classics, and asked me which books could be translated, I suggested Mencius, and this time it took me seven years to translate the book… ..[32].

To not misinterpret Lao Tzu’s philosophical thoughts, D.C.Lau decided to translate Chinese classics by himself with a rigorous and serious attitude after refusing the publication of two translations. Later, D. C. Lau translated Mencius. His main purpose was to introduce the essence of Chinese philosophy to the readers in the target language world more accurately and completely. Therefore, he adopted the foreignization translation strategy and the “thick translation” method and used a variety of auxiliary text materials to interpret the translation. Deng [38] once commented on the translation of D.C.Lau’s Mencius:

Professor Liu has the academic foundation of Qianjia period since childhood, and especially loves the exegetic knowledge of the Er Wang. Coupled with the modern Western academic training, he is not limited to knowledge of one school, his insights and propositions held have a certain basis, and the harvest of classic research can be said to be beyond the scholars of Qianjia period. The translation is rooted in deep learning, and there is deep insights in the judgement of every word and sentence [38].

To some extent, scholars’ high evaluation of D.C.Lau’s translation reflected the features of rigorous and accurate words in the translation. However, Chen [39] criticized the Lao Tzu translation for not specifying annotations or other documents supporting his translation. In response to this criticism, Liu responded as follows:

The target audience of my translation is mainly the general public readers, so there is no need to trace the source in the annotation, otherwise it can only arouse the interest of sinologists in the West, not the general public. And I translated Chinese Classics at the request of Penguin Publishing, which is one of the largest publishers of popular books in the world, and its “Penguin Classics” series is characterized by common readers’ popularity [32].

His translation showed the simple style of Mencius and expressed his profound thoughts. D.C.Lau has a strong reader awareness. Although there are various notes in his translation, he does not excessively present too much explanation and rationale and leaves the reader a broader space for thinking. Moreover, to eliminate the barriers between Chinese and English, and help readers understand the content outside the text, the translation contained 40 pages of introduction and six appendices. In the introduction, Mencius’ life and thoughts are emphatically introduced, and the connotation of some important thoughts is explained in detail. His translation strategy is conducive to the wide dissemination of the translation in the West. This will greatly help Western readers living in different cultural backgrounds to understand Mencius’ thoughts from all angles. For his translation,Yang [40] thinks that “there are short sentences, concise and symmetrical structure in D.C.Lau’s translation, which reflects the oral dialog characteristics of the original text and is easy to understand.”