《经济学人》科技类新闻机器翻译译后分析

Chapter 1 Introduction

1.1 Background of the Translation Practice

In the globalized society with the rapid development of economy and trade all over theworld, language communication is an important factor to promote the good development ofthis society. Therefore, the level of the global language service industry is also required. The“MT + PE” (Machine translation + Post-editing) model is the product bred under thisbackground. Giant companies such as SDL, Microsoft and Google have established their own

professional post-editing teams, and at the same time, other language service providers in theindustry are also providing post-editing related services. The European Language ServiceIndustry Association conducted a professional survey on its language service providers in2018. The survey shows that 42% of the companies have post-editing related business lines,22.2% of the companies also specially train professional post-editors, 72% of the companieswill have a set of programs to deal with machine translation materials, while furtherpost-editing will be carried out on the translation. (ELSIA, 2018). At the same time, 57.1% ofthe companies take post-editing business as a new opportunity for the development of thecompany and intend to further develop and improve it (Feng Quangong, 2019). CommonSense Advisory conducted a survey and evaluation on the global language service market in2018. The survey result shows that the post-editing service of machine translation accountsfor 3.94% of the global language service market. Among them, 25.9% of the language serviceproviders have specialized post-editing business and believe that post-editing will have abroader market in the future (CSA, 2018). At the same time, TAUS, SDL, Google and otherlanguage services related institutions have already started to provide post-editing trainingprograms. With the increasing improvement of machine translation and the increasingdemand of professional translation, the cost of manual translation is accordingly increasing. Inthis environment, post-editing mode will become the mainstream of language serviceindustry

1.2 Characteristics of Scientific and Technological Text

In order to do a good job in the translation of scientific and technological text, the mostimportant thing is to learn the unique characteristics of scientific and technological text.Actually, each language has three different characteristics: Each language has its own uniquestructure; each language can express specific practical activities by way of elaboration; andeach language has its specific development environment. English and Chinese belong todifferent language systems. English is one of brands related to Indo-European languagefamily, but Chinese comes from Sino-Tibetan language family. Therefore, a lot of differencesappeared in the structure and style of writing between English and Chinese. English is alanguage that emphasizes sentence structure and form. Multiple repeated sentences canexpress more rigorous and detailed complex activities than simple sentences. Therefore, it isconsidered to be an important syntactic phenomenon that can best reflect the characteristics ofscientific and technological text (Fan Wuqiu, 2011:166).

Chapter 2 Literature Review

2.1 Previous Studies on Error Types of Machine Translation

In today’s society with the rapid development of science and technology, machinetranslation technology is also developing to meet the needs of people for a large number oftext translation. During this period, scholars have done a lot of research, and made a series ofchanges to improve the performance of machine translation system.Through CNKI’s literature retrieval method, with the subject of “Machine TranslationError Analysis”, 35 papers were retrieved. The first related literature appeared in 2001, andthe number of literatures has increased year by year after 2012. From the above data, it can befound that although the research on machine translation error analysis in China startedrelatively late and the number of related research papers is relatively small, the research trendis showing an increasing trend. After sorting out the relevant literature collected, it is foundthat the research on machine translation error analysis is mainly focused on the followingaspects: (1) There are research on the typical error forms in translation processing such asvocabulary and symbols. (2) There are research on the errors of short sentences and clauses inmachine translation. (3) There are some studies on the typical error forms of different texttypesthose at the syntactic level, they cannot be ignored (Li Mei & Zhu Ximing, 2013:207).

2.2 Previous Studies on Post-Editing of Machine Translation

Through CNKI’s literature retrieval method, with “Post-editing” as the subject, 408literature were retrieved. The first related literature appeared in 1984 and it is a foreignlanguage literature, and the number of literatures has increased year by year after 1990s.Among them, the early literature is basically foreign literature, and the domestic researchbasically began gradually after 2010. From the above data, it can be found that due to thedevelopment of science and technology, machine translation basically developed abroad, andthe research on post-editing strategies is also earlier than that in China for a long time.Although the research on post-editing strategies in China starts relatively late and the numberof related papers is relatively small, the research trend is showing an increasing trend. Thecontent of the research on post-editing is also different in different stages.

Chapter 3 Description of the Translation Process........................................................... 9

3.1 Preparation for Translation........................................................................................................... 9

3.1.1 Selection of Translation Tools...................................................................................................... 9

3.1.2 Operational Principles for Post-Editing......................................................................................10

3.2 During the Translation................................................................................................................. 11

3.2.1 Term Base Establishment............................................................................................................11

3.2.2 Quality Control............................................................................................................................11

3.3 After the Translation.................................................................................................................... 12

3.3.1 Translation Inspection.................................................................................................................12

3.3.2 Sorting Out the Corpus................................................................................................................13

3.3.3 Writing Arrangement.................................................................................................................. 13

Chapter 4 Typical Errors of Machine Translation in This Translation Practice............................14

4.1 Mistranslation of Words..............................................................................................................14

4.1.1 Errors in Selecting the Wrong Meaning..................................................................................... 14

4.1.2 Mistranslation of Terminology................................................................................................... 15

4.1.3 Mistranslation of Conjunctions...................................................................................................15

4.1.4 Omission in Translation.............................................................................................................. 16

4.2 Mistranslation of Phrases............................................................................................................ 17

4.2.1 Mistranslation of Verb Phrases...................................................................................................17

4.2.2 Mistranslation of Noun Phrases.................................................................................................. 17

4.2.3 Mistranslation of Prepositional Phrases......................................................................................18

4.2.4 Mistranslation of Parenthesis......................................................................................................18

4.3 Mistranslation of Clauses............................................................................................................19

4.3.1 Mistranslation of Attributive Clauses......................................................................................... 19

4.3.2 Mistranslation of Adverbial Clauses...........................................................................................19

4.3.3 Mistranslation of Appositive Clauses......................................................................................... 20

4.4 Mistranslation of Unity and Format......................................................................................... 20

4.4.1 Mistranslation of Proper Nouns in its Unity...............................................................................20

4.4.2 Mistranslation of Punctuations....................................................................................................21

Chapter 5 Post-Editing Strategies for Errors of Machine Translation............22

5.1 Corpus-Oriented Strategy for Mistranslation of Words.......................................................22

5.2 Adjusting the Order for Mistranslation of Phrases................................................................23

5.3 Manually Dividing the Sentence for Mistranslation of Clauses........................................ 25

5.4 Using CAT Tools for Mistranslation of Unity and Format.................................................27

Chapter 6 Conclusion....................................................................................................................... 29

6.1 Machine Translation’s High Accuracy in Semantic Translation.......................................29

6.2 Unavoidable Problems of Machine Translation.................................................................... 30

6.3 Gains in the Translation Practice...............................................................................................30

References.................................................................................................................................................32

Acknowledgments................................................................................................................................34

Chapter 6 Conclusion

6.1 Machine Translation’s High Accuracy in Semantic Translation

Due to the limitation of the principle of machine translation, the machine usually givespriority to word-by-word translation when recognizing the input text. If the phrase or sentencecan find the corresponding reference translation in the machine translation corpus, then themachine can use the translated corpus to deal with the input texts. If the reevant referencematerial is not retrieved in the corpus, the machine will tend to translate the input texts wordby word. As mentioned at the beginning of the report, the characteristics of scientific andtechnological text are concise and clear, and the sentences are simple and easy to understand.Therefore, the machine has no burden to recognize the original text most of the time, and themachine can accurately translate the semantics of the sentence. This is also the presence ofMT output and the PE result is almost exactly the same, because the quality of machinetranslation is already very high, there is no need to do extra polishing work, which violates thestarting point of using machine translation to pursue efficiency.Looking closely at the translation itself, machine translation has performed well intranslating most professional vocabulary. This is mainly because a large part of scientific andtechnological text is aimed at the public, although this is also part of the terminology, thepublic has already been familiar with these terms, and the relevant corpus is also rich. Somachine translation can show a good performance with the support of the corpus. At the levelof phrases and clauses, due to the popular and straightforward nature of scientific andtechnological text, the phrases and sentences used in the source text are relatively simple andeasy to understand, and there are no particularly many modifiers, so the machine can translatemore accurately during the process of recognition and translation and can translate thesemantics of the source text well. At the unity and format level, the symbol match and theinconsistency of words are the common problems that occurs in many different texts, which isnot the privilege for scientific and technological text only, so this part is also a translation thatrequires post-editing work.