论文总字数：42545字

摘 要

国内外许多翻译家和文学批评家对于译者风格已经探讨了很长一段时间。近年来，随着计算机技术水平的不断提高，一些翻译理论家开始将语料库应用于翻译理论研究，对译者风格做定量研究，对译者语言做统计分析，从而更科学地得出译者的翻译风格。

本论文将首先介绍语料库翻译学的理论基础及研究现状，接着以麦尔维尔的小说《白鲸》为例，选取成时和曹庸两个中译本，利用语料库的研究方法在词汇层面上，通过比较词频概况、词汇密度以及标准类符形符比这三个方面的差异，对这两个译本进行定量和定性的对比分析；同时，结合两位译者的相关生活经历比较，从而总结出两位译者的翻译风格方面的差异；最终对全文进行总结，并指出研究的局限与不足。

关键词：译者风格；语料库翻译学；Moby Dick

Table of Contents

Acknowledgements i

Abstract ii

摘要 iii

Table of Contents iv

Chapter One Introduction 1

1.1 Background of the study 1

1.2 Significance and purpose of the study 1

1.3 Layout of the thesis 2

Chapter Two Literature Review 3

2.1 Definitions 3

2.2 Current research based on corpus translation studies 3

Chapter Three The Research Process 6

3.1 Introduction of Cheng Shi’s and Cao Yong’s E-C translations 6

3.2 Corpus Building 7

3.3 Research tool 8

3.3.1 AntConc 3.5.7 8

3.3.2 ICTCLAS 2015 9

3.3.3 CLAWS 7 9

3.4 Reference Corpus 10

Chapter Four A CTS-based Study on Two E-C Translated Versions of Moby Dick 11

4.1 Lexical features of the two translated versions……………….....…………..….11

4.1.1 Word frequency…………………………………...………………………………11

4.1.2 Lexical density………………………………………………………………...…..13

4.1.3 Standard Type-token ratio (STTR)………...…………………………………...…15

4.2 Analysis of the differences between the two translators’ styles………...…..…17

4.2.1 Life experience……………………….…………...………………………………17

4.2.2 Translation thoughts………………….…………...………………………………18

Chapter Five Conclusion 19

5.1 Summary…………………………...................……………………………….19

5.2 Limitations and suggestions for further study…………………………......…..19

References 20

Chapter One Introduction

1.1 Background of the study

Translator’s style has been discussed by many domestic and foreign translators and literary critics for quite a long time. As a result, it is not a new subject. However, the previous studies on translator’s style were mainly based on impression and intuition to judge whether the original translating style is persisted or diverted to its translation. This kind of method usually lacks scientific evidence and cannot convince most people.

Fortunately, with the development of computer technology, corpus linguistics has made much progress. Some scholars began to combine Corpus Linguistics with Descriptive Translation Studies, thus contributing to the appearance of Corpus Translation Studies (CTS). Professor Mona Baker (2000) first made use of corpus to study the differences between the translating styles of Peter Bush and Peter Clark. This encouraged many scholars to focus on the study of translator’s style with the help of corpus.

Moby Dick is an outstanding novel written by Herman Melville in 1851.Given that only a few scholars adopted corpus software to study the translation of Moby Dick, the author will try to analyze the translator’s style of Moby Dick based on corpus.

1.2 Significance and purpose of the study

Moby Dick is an excellent novel written by Melville. It has been translated into many different languages and introduced to many foreigners. However, quite a few scholars focus on the comparison of the different translated versions of the novel. Even if some people do it, most of them tend to choose the qualitative methods, which usually lack convincing evidence. This thesis is significant in the following aspects: First, it combines qualitative methods with quantitative methods from the level of corpus, which is a brand new study on the translations of the novel. Second, it can help the readers choose a more suitable version of Moby Dick when they don’t know which one is better for them.

This thesis attempts to make a comparison between the two Chinese versions of Moby Dick created by Cheng Shi and Cao Yong from the lexical level with the aid of corpus software. This study is expected to make a quantitative and qualitative analysis of the two translations, in order to conclude the differences between the two translators’ styles and promote the development of Corpus Translation Studies.

1.3 Layout of the thesis

The structure of the thesis consists of five chapters. Chapter One is a general introduction of the thesis, including the background, significance and purpose of the study and the layout of the thesis. Chapter Two is the literature review, which includes the definitions and the current research on translators’ styles on the basis of corpus. Chapter Three is the research process. This chapter covers the introduction of Cheng Shi’s and Cao Yong’s E-C translations and the corpus building. Also, the research tool (Antconc, ICTCLAS and CLAWS) and the reference corpus (LCMC) will also be mentioned. Chapter Four is a CTS-based study on two E-C translated versions of Moby Dick. It includes an analysis of the lexical features of the translator’s styles generally from the aspect of word frequency, lexical density and the standard type-token ratio (STTR). In addition, the thesis takes the life experience and translation thoughts of the two translators into consideration and then makes an analysis of the differences between the two translators’ styles. The last chapter, Chapter Five, makes a summary of the above study, points out the limitations and gives some suggestions for future study.

Chapter Two Literature Review

2.1 Definitions

Translator’s style and corpus translation studies are the basic framework of the study.

Hermans (1996) firstly put forward the concept of translator’s voice in The Translator’s voice in Translated Narrative. He believed that translator’s voice can appear everywhere in the translation. In his opinion, translator’s style is the translating methods adopted by translators. Then, Mona Baker gave a more concrete definition for translator’s style. She argued that translator’s style is “a kind of thumbprint that is expressed in a range of linguistic--as well as non-linguistic—features” (Baker 2000). As for Baker, a study of a translator’s style must focus on the manner of expression that is typical of a translator, rather than simply instances of open intervention. It must attempt to capture the translator’s characteristic use of language and individual profile of linguistic habits.

Corpus translation studies refers to the studies that systematically analyzes the translation process, translation phenomenon and translation essence based on corpus and the linguistics, literature and translation theory by taking the real bilingual text or translated text as the research object, adopting the research method of data statistics and theoretical analysis (Baker 1993). It was a subject created by Professor Mona Baker in her paper Corpus Linguistics and Translation Studies: Implications and Applications in 1993. Corpus Translation Studies gradually became a significant subject in the study of translation owing to its systematic and scientific research methods since 1990s.

2.2 Current research based on corpus translation studies

Since 1990s, with the great development of computer technology, corpus linguistics has made much progress. More and more scholars and literary critics began to apply it into descriptive translation studies, thus creating a new subject, which is Corpus Translation Studies.

It is commonly considered that the first one to study translator’s style with the aid of corpus is Professor Mona Baker. In 2000, she made a comparison between the translator’s style of Peter Bush and Peter Clark. She emphasized that it was the different difficulty of the original article, life experiences and translating methods that result in their different translating styles. Maeve Olohan (2003) also analyzed the use of complete form and its abbreviation in the two translators’ translations. The study showed that Peter tended to use abbreviations, while Peter Clark prefers the complete form. The difference in their translating styles originated from their different language habits.

Charlotte Bosseaux (2006) compared and analyzed two French translations of Virginia Woolf’s work The Waves from the aspect of type-token ratio, mean length of utterance and cultural lexicons. The result revealed that one translator tried to keep the original foreign culture in the translation while the other one wanted to eliminate it.

What’s more, Meng Ji (2009) from London University also created a parallel corpus of Don Quixote written by a Spanish writer, Saavedra. Then, he made a systematical analysis of the application of four-word idioms in the translations of Don Quixote by Yang Jiang and Liu Jingsheng. The study pointed out that Yang preferred to use reduplications while Liu tended to use figurative idioms.

These are some examples of current studies on translator’s style based on corpus abroad. Nowadays, studies at home also have made rapid progress.

Liu Zequan, Liu Chaopeng and Zhu Hong (2011) analyzed the differences between four translated versions of Hong Lou Meng mainly from the lexical and syntactical level with the aid of the parallel corpus built by themselves. The result showed that every translated version had their characteristics. Zheng’s version was quite easier to understand while Qiao’s version seemed more complex. Huo’s version was translated to a narrative English one, which may be closer to western culture. Yang’s version was the hardest for foreigners because it kept the most domestic culture shown in the novel.

Xu Xin (2010) compared the three Chinese versions of Jane Austen’s novel, Pride and Prejudice with the aid of corpus. He mainly focused on three aspects, including the type-token ratio, high frequency words and idiosyncratic words. The study finally showed that Sun Zhili’s version has the highest type-token ratio. Wang Qing and Liu Li (2014) studied the formalization of vocabulary in the Chinese version of Ulysses with the aid of corpus. Liu Xiaoyao (2017) also compared the two Chinese versions of Moment in Peking from the aspects of vocabulary and sentence. What’s more, Zhang Dandan and Liu Zequan (2014) carried out a research on the translation of A Dream in Red Mansions with the help of corpus to study whether it was created by one translator.

The above all show that studies on translator’s styles based on corpus has improved rapidly due to the development of computer technology. More and more scholars will pay attention to the field of corpus translation studies and translator’s styles.

Chapter Three The Research Process

3.1 Introduction of Cheng Shi’s and Cao Yong’s E-C translations

Moby Dick is generally considered as a best-seller all over the world. It is also known as The Whale, a novel written by Herman Melville, first published in 1851. It is considered to be one of the greatest American novels and a treasure of world literature. The story tells the adventures of a wandering sailor Ishmael, and his voyage on the whaleship Pequod, commanded by Captain Ahab. The enormous white whale called Moby Dick torments Captain Ahab, who is obsessed with finding and killing Moby Dick, having lost a leg in a previous encounter with the wale. Captain Ahab’s burning desire for revenge really is the center of the story. In the end, Ahab finds and attacks Moby Dick. On the first day the wale overturns a boat. Next day, it swamps another. When the third day comes, Ahab and his crew manage to plunge a harpoon into it, but the wale carries the Pequod along with it to its doom. All on board the whaler get drowned, except one, Ishmael, who survives to tell the tale. Moby Dick is ruthless in attacking the sailors who attempt to hunt and kill him, but it is Ahab who invests Moby Dick’s natural instincts with malignant and evil intentions. In fact, it is not the whale but the crippled Ahab who alone possesses this characteristic.

This novel has been translated into different languages, including many Chinese versions. Different Chinese versions all show different translating styles of the corresponding translators. In order to make the result of the study more scientific and convincing, we searched online and read different Chinese versions of Moby Dick in advance. Finally, Cao Yong’s version and Cheng Shi’s version were chosen for a comparison.

Firstly, these two translators are commonly regarded as famous and respectable translators in China. Cao Yong, whose original name was Hu Hanliang, worked in Shanghai Translation Publishing House as an editor for a long time. He started translating literary works from 1937 and translated quite many famous English novels, including Cat in the Rain, Moby Dick and The Killers. He began the translation of Moby Dick in June, 1982. His Chinese version of Moby Dick was firstly published by Shanghai Translation Publishing House in 1990. Till now, his version has been published for many times by different publishing houses. The newest version of Cao Yong’s version published by Shanghai Translation Publishing House was in 2008. As for Cheng Shi, his original name was Xu Chengshi. He was a new editor in Xinhua News Agency in 1951. In 1984, he joined the Chinese Writers Association. He has translated a lot of Russian novels and American novels throughout his whole career. His English translations included The Adventures of Huckleberry Finn, The Adventures of Tom Sawyer and Moby Dick. Cheng Shi’s translation of Moby Dick was firstly published by People’s Literature Publishing House in 2001. After that, the publishing house also published Cheng’s version for a great many times. The latest version of it was published in 2011.

Secondly, the two translations are said to be different from each other. It may have something to do with a lot of things, including the two translators’ life experiences, translating habits. This kind of difference can give the readers more space to choose a more suitable version for them to appreciate. This is also the reason for the author to choose these two translations to make a comparison.

3.2 Corpus Building

The original text of the two E-C translations are in PDF. As a result, the first step is to convert them into TXT form. Here I make use of CAJ viewers to successfully change them into TXT versions. After that, the CLAWS 7 and ICTCLAS 2015 were used to do tokenization and text annotation. The following pictures are part of the self-build corpus:

3.3 Research tool

Before carrying out the research, the researcher must have a basic knowledge of the related software to be used during the research in order to get the essential data for further study and analysis. As a result, the following part will introduce the specific software in detail.

The software used in the study include: AntConc 3.5.7, ICTCLAS 2015 and CLAWS 7. We also got the PDF version of the two translations and transferred them into TXT files with the help of CAJviewer 7.2. However, this process was just a manual work costing time. It is definitely not difficult at all. As a result, there’s no need to further introduce the detailed usage of CAJviewer 7.2.

First of all, CAJviewer 7.2 was used to transfer the PDF version of the two translations into TXT files. Then ICTCLAS 2015 and CLAWS 7 were exploited to carry out word segmentation. After finishing segmenting the words and detecting errors, we used AntConc 3.5.7 to analyze the related statistics such as word frequency, lexical density and type-token ratio.

3.3.1 AntConc 3.5.7

AntConc Tool is a free software created by Laurence Anthony, a professor from Center for English Language Education in Science and Engineering, School of Science and Engineering, Waseda University. Laurence designed this software for the researchers to do research on corpus linguistics and data-collecting. This software can be run on any computer with systems including Microsoft Windows, Macintosh OS X and Linux. Due to its low cost and extensive compatibility, more and more scholars and researchers begin to apply it to their research.

AntConc has been constantly improved and upgraded by the designer. It has changed greatly from the first 1.0 version to the latest 3.5.7 version. The latest version mainly has seven tools, which are the Concordance Tool, Concordance Plot Tool, File View Tool, Clusters Tool, Collocates, Word List and Keyword List. Here is the introduction of three main tools used in this thesis.

The first one is Word List. It is usually used to count the frequency of every word and make them in an order list. After that, the researcher can use this tool to get the high frequency words used in the related translations.

The second tool is Keyword List, which can help identify the key words used on the corpus, so that the researcher can compare them with those in the reference corpus.

The third one is the File View Tool. It can show the content of every single file for the researchers to see the results when they are using other tools of AntConc.

3.3.2 ICTCLAS 2015

ICTCLAS 2015 is a Chinese Lexical Analysis System designed by Institute of Computing Technology. It is also one of the most efficient software to make lexical analysis of Chinese words.

The main function of the software includes Chinese Segment, annotation, named entity recognition, new words discovery and keyword extraction. The system can also distinguish traditional Chinese. Apart from this, it is equipped with user dictionary for a more comfortable user experience. The segmenting speed of the software is up to 1048KB/s and the segmenting accuracy is 99.12%. What’s more, in order to meet the demand of different researchers, the system can also be designed into a special segmenting system suitable for specific researchers.

3.3.3 CLAWS 7

CLAWS, also known as the Constituent Likelihood Automatic Word-tagging System, is used to POS tag millions of words of the British National Corpus. The precise degree of accuracy of CLAWS can reach up to nearly 97%. However, the whole system only has 1.5% of error-rate and 3.3% of ambiguities.

In the context of the BNC Enhancement project, University Centre for Computer Corpus Research on Language (UCREL) devised a Template Tagger to act as a post processor for CLAWS. The rule-based formalism implemented in the Template Tagger is more powerful than that built into CLAWS itself. Manual corpus analysis and knowledge of frequent CLAWS tagging errors was used to create a rule base for the tool. This facilitated an improvement in the tagging accuracy in the resulting corpus.

3.4 Reference Corpus

In order to set a standard of comparison, researchers usually take advantage of a reference corpus. This kind of reference corpus is regarded as a benchmark. As a result, it must be convincible.

The thesis uses the famous LCMC corpus as a reference corpus. The LCMC corpus is also known as the Lancaster Corpus of Mandarin Chinese. It is a one-million-word balanced corpus of written Chinese created by Professor Xiao Zhonghua. This corpus is part of the achievement of a project called Contrasting Tense and Aspect in English and Chinese funded by the British Economic and Social Research Committee.

The LCMC corpus is marked up in XML format at five levels: text category, sample file, paragraph, sentence and token. Professor Xiao also made manual proofreading on the texts to make sure the accuracy of the part-of-speech tagging can be up to 98% or more. The corpus can be freely used by any scholar or researcher for academic research or education without commercial purposes. However, researchers should also be aware of the limitations of LCMC when doing a research. Except for these tiny limitations, LCMC is still one of the most reliable and scientific corpus for research or education.

Chapter Four A CTS-based Study on Two E-C Translated Versions of Moby Dick

4.1 Lexical features of the two translated versions

Before making a comparison from the specific three aspects, a brief introduction of the two Chinese versions of Moby Dick is presented. Here is the table:

Table 1.1 Brief Introduction of Cao Yong’s and Cheng Shi’s Translations

With the aid of corpus software, the two translated versions will be compared on the lexical features. To be specific, the thesis will analyze the two translations mainly from three aspects: word frequency, lexical density and standard type-token ratio (STTR). The LCMC corpus will also be used as a reference corpus to show the differences between the two translators’ styles. Apart from that, the author will give some specific examples of the two translations to prove them.

4.1.1 Word frequency

Word frequency refers to the frequency that a word appears in a sentence, an article or a book. Those words that appear in a text for quite many times are referred to as high frequency words. As we know, different translators may have different life experiences, language habits and literature accomplishments. All of these things may influence the words used in their translation works.

With the help of AntConc 3.5.7, the author got the frequency of the words used in Cao’s version and Cheng’s version. To make the result much easier to be understood, the author chose the top ten high frequency words appeared in Cao’s version and Cheng’s version. In addition, the top ten high frequency words in LCMC corpus are also collected together. Here the author made a table of the high frequency words:

Table 1.2 Top Ten High Frequency Words of the Two Translations and LCMC

According to the table, we can see that the Top Ten high frequency words are different in order. However, there are still many high frequency words appearing together in all the three origins.

We can find from the table that the functional word “的” comes first all in the two Chinese versions of Moby Dick and the LCMC corpus. Apart from this, pronouns like “我” and “他” also occupy significant status in the table. This also shows that pronouns are widely used in Chinese texts, especially novels which focus on the plot and the characters.

What’s more, we can see that the word “那” appears in the high frequency words of Cao’s version instead of Cheng’s version. The interjected word “了” comes up as a high frequency word in Cheng’s version while disappears in Cao’s version. To be specific, we can find that “那” was often used in the traditional Chinese language, which seems to be more formal and uneasy to understand. On the contrary, the word “了” tends to be used more frequently in contemporary Chinese language to express the abundant emotions of the characters, which is much more humorous and flexible than Cao’s version. Compared with the high frequency words in LCMC, we can find that the high frequency words in the two translations are basically similar to those in LCMC.

4.1.2 Lexical density

According to Professor Mona Baker, lexical density refers to the percentage of lexical as opposed to grammatical items in a given text or corpus of texts. Then here comes a question. What is lexical words or grammatical words?

Lexical words usually include nouns, verbs, adjectives and adverbs. On the contrary, grammatical words are those words with grammatical meanings, including conjunctions, prepositions, articles, numerals and pronouns. Grammatical words are also referred to as functional words. The density of a speech or text is calculated by comparing the number of lexical words and the number of functional words. Balanced lexical density is approximately 50 percent. This means that half of each sentence is made up of lexical words and half of functional words. A low-density text will have less than a 50:50 ratio and a high-density text will have more than 50:50. Academic texts and government, jargon-filled documents tend to produce the highest densities.

Apart from that, Professor Mona Baker pointed out that the lexical density of translated English is commonly lower than that of original English. The following table is about the lexical density of Cao’s version and Cheng’s version of Moby Dick.

Table 1.3 Lexical Density of the Two Translations of Moby Dick and LCMC

As we can see from the table, the lexical density of Cao’s version is 63.12% while the lexical density of Cheng’s version is 61.47%. It is commonly considered that a text with a higher lexical density is usually more formal. This means that the related text has a larger percentage of content words with a smaller gap between every single word. This means that the text will be more difficult for readers to understand. What’s more, the lexical density in Cheng Shi’s version is closer to that in LCMC, which means that Cheng Shi’s version is closer to the Chinese narrative way.

Example:

Moby Dick: Call me Ishmael. Some years ago——never mind how long precisely——having little or no money in my purse, and nothing particular to interest me on shore, I thought I would sail about a little and see the watery part of the world. (page 1)

Cao’s version: 管/v 我/r 叫/v 以实玛利/nh 吧/u 。/w 几年/m 前/p ——/w 别/d 管/v 它/r 究竟/a 是/v 多少/m 年/n ——/w 我/r 的/u 荷包/n 里/p 只/a 有/vl 一点点/m ,/w 也/a 可以/p 说/v 是/v 没有/vl 钱/n ,/w 岸上/n 也/a 没有/v 什么/a 特别/a 让/v 我/r 留恋/v 的/u 事情/n ，/w 我/r 想/v 我/r 还是/v 出去/v 航行/v 一番/m ，/w 去/v 见识/v 见识/v 这个/r 世界/n 的/u 海洋/n 部分/n 吧/u 。/w (page 1)

Cheng’s version: 你/r 就/p 叫/v 我/r 以实玛利/nh 吧/u 。/w 那/r 是/v 有/v 些/m 年头/n 的/u 事/n 了/u ——/w 到底/a 是/v 多少/m 年/n 以前/a ，/w 且/a 不去/v 管v 它/r ——/w 当时/a 我/r 口袋/n 里/p 没有/vl 几个/m 钱/n ，/w 说/v 一文/n 不名/n 也/a 未尝不可/a ，/w 而/a 在/p 岸上/n 又/a 没有/v 特别/a 让/v 我/r 感兴趣/v 的/u 事/n 可干/v 。/w 我/r 于是/a 想/v ，/w 不如/v 去/v 当/v 一阵子/m 水手/n ，/w 好/a 见识/v 见识/v 那/r 水/n 的/u 世界/n 。/w (page 2)

From the above example, we can see that there are 52 tokens in the exerted part of Cao’s version. For Cheng’s version, it has 62 tokens. Then, we can take a look at the content words in the two E-C translations. Cao’s version has 33 content words in this part, while Cheng’s version has 32 content words. This shows that Cao’s version has a higher lexical density than Cheng’s version. What’s more, we can find from the text that Cao’s version carries many old words like “管”, “究竟” and “荷包”. Also, Cao’s version has a smaller gap between words.

When it comes to a translation work, it means that this translation carried more things or elements from the original literary work. As a result, the translator may show more respect for the culture hidden in the original work. As a matter of fact, this kind of translation may employ the method of literal translation. So some foreign readers may find it difficult to fully understand the meaning of the idea or culture conveyed by the author. In other words, Cao’s version of Moby Dick may be more difficult for the readers to understand due to its high respect for the original Moby Dick.

4.1.3 Standard Type-token ratio (STTR)

Standard Type-token ratio (STTR) refers to the average ratio of different words to the sum of total words in texts. This ratio is generally used to measure the vocabulary feature or diversity of a translator or writer. The author took advantage of the AntConc 3.5.7 software to make a comparison between Cao’s and Cheng’s version of Moby Dick. The LCMC corpus was also used as a reference. The detailed data are shown in the following table:

Table 1.4 Types, Tokens and TTR of the Two Translations and LCMC

From the table, we can see that Cao’s version has 168,750 tokens, while Cheng’s version has 182,083 tokens. The tokens in Cao’s version is much fewer than those in Cheng’s version. However, we can also see that the types in Cao’s version is more than those in Cheng’s version. As readers who have read the two translations will easily find that Cheng Shi preferred to use more grammatical or functional words to make the content of the translation work much easier for readers to understand. What’s more, According to the statistics, the standard type-token ratio (STTR) in Cheng Shi’s version is closer to that in LCMC than the STTR in Cao Yong’s version. This also helps to prove that Cheng Shi’s version is closer to the Chinese culture from the lexical level.

剩余内容已隐藏，请支付后下载全文，论文总字数：42545字