Machine Trans EN 13

Machine Translation - A challenge or a chance for human translators?

Overview Page of Machine Translation

30 Chapters（0/30)

Machine_Trans_EN_1 Machine_Trans_EN_2 Machine_Trans_EN_3 Machine_Trans_EN_4 Machine_Trans_EN_5 Machine_Trans_EN_6 Machine_Trans_EN_7 Machine_Trans_EN_8 Machine_Trans_EN_9 Machine_Trans_EN_10 Machine_Trans_EN_11 Machine_Trans_EN_12 Machine_Trans_EN_13 Machine_Trans_EN_14 Machine_Trans_EN_15 Machine_Trans_EN_16 Machine_Trans_EN_17 Machine_Trans_EN_18 Machine_Trans_EN_19 Machine_Trans_EN_20 Machine_Trans_EN_21 Machine_Trans_EN_22 Machine_Trans_EN_23 Machine_Trans_EN_24 Machine_Trans_EN_25 Machine_Trans_EN_26 Machine_Trans_EN_27 Machine_Trans_EN_28 Machine_Trans_EN_29 Machine_Trans_EN_30 ...

Back to translation project overview

To the To Do list 13 陈湘琼Chen Xiangqiong（Study on Post-editing from the Perspective of Functional Equivalence Theory ）

Abstract

With the development of technology，machine translation methods are changing. From rule-based methods to corpus-based methods，and then to neural network translation，every time machine translation become more precise, which means it is not impossible the complete replacement of human translation by machine translation. But machine translation still faces many problems until today such as : fail to translate special terms, incapable to set the right sentence order, unable to understand content and culture background etc. All of these need to be checked out and modified by human translator, so it can be predict that the model Human + Machine will last for a long period. This article will discuss mistakes made in machine translation and describe what translators should do in post-editing based on the skopos theory and functional equivalence theory

Key words

machine translation，post-editing，skopos theory，functional equivalence theory

题目

基于功能对等视角探讨译后编辑问题与对策

摘要

随着科技的不断发展，机器翻译方法也在不断变革，从基于规则的机器翻译，到基于统计的机器翻译，再到今天基于人工神经网络的机器翻译，每一次变化都让机器翻译变得更精确，更高质。这意味着在不远的将来，机器翻译完全代替人工翻译成为一种可能。但是直至今天，机器翻译仍然面临许多的问题如：无法准确翻译术语、无法正确排列句子语序、无法分辨语境等，这些问题依然需要人工检查和修改。机器翻译自有其优点，人工翻译也有无可替代之处，所以在很长一段时间内，翻译都应该是机器+人工的运作方式。本文将基于翻译目的论和功能对等理论，对机器翻译可能出现的错误之处进行探讨，并且旨在描述译者在进行译后编辑时需要注重的方面，为广大译员提供参考。

关键词

机器翻译，译后编辑，翻译目的论，功能对等

1. Introduction

For a long time, researchers believe MT may have seemed relatively peripheral, with limited use. But recently, because of the technological advances in the field of machine translation, the translation industry has been experiencing a great revolution where the speed and amount of translation has been raised desperately. So, the idea that human translation may be completely replaced by machine translation in the future may come true. This changing landscape of the translation industry raises questions to translators. On the one hand, they earnestly want to identify their own role in translation field and confront a serious problem that they may lost job in the future. On the other hand, in more professional contexts, machine translation still can’t overcome difficulties such as: fail to translate special terms, incapable to set the right sentence order, unable to understand content and culture background etc. For this reason, human-machine interaction is certainly becoming a trend in the recent future. Therefore, translators start to use machine translations as raw versions to be further post-edited, which becomes the topic we want to discuss today. This paper presents a research investigating the post-editing work in machine translation. From the prospect of functional equivalence and skopos theory, we discuss the errors machine translation may made in the process and what strategies translator should use when translating. Section 2 provides an overview of the two theories and the development in the practical use. Section 3 presents debates on relationship between MT and HT. Section 4 review the history and development of post-editing.

2. Functional Equivalence and Skopos Theory

Functional equivalence theory is the core of Eugene Nida’s translation theory, who is a famous translator and researcher in America. It aims to set a general standard for evaluating the quality of translation. In his theory Nida points out that “translation is to convey the information from source language to target language with the most proper and natural language.”(Guo Jianzhong, 2000:65) He holds that translator should not only achieve the information equivalence in lexical sense but also take into account the cultural background of the target language and achieve the equivalence in semantics, style and literature form. So the dynamic equivalence contains four aspects: 1. lexical equivalence；2.syntactic equivalence；3.textual equivalence；4.stylistic equivalence, which basically construct and guide the idea of this article.

In 1978, Hans Vermeer put forward skopos theory in his book Framework for a General Translation Theory. In this theory, he believes that translation is a human activity which means it has special purpose in itself like other human activities.(Nord, 2001:12) Also, there are some rules that the translator should follow in the progress of translation: 1.purpose principle; 2.intra-textual coherence; 3.fidelity rule, which exactly shows its correlation with machine translation.

According to these two theories, we can start now to explore some principles and standard that translator ought to obey in post-editing. Firstly, efficiency and accuracy are really important because the translator’s purpose is obviously raising money in comparatively short time. If they fail to provide translation with high quality or if they unable to finish the job before deadline, the consequence will be relatively bad. Secondly, translators need to achieve equivalence in lexical level, syntactic level, textual level and stylistic level in post-editing for the reason that machine translation can be always misunderstood when they are dealing with words and sentences with special background knowledge. Thirdly, it is almost impossible for machine translation to achieve communicative goal and fulfil cultural exchange that human brain is indispensable to jump over the gap. And more details will be discussed later on.

3. Machine Translation Versus Human Translation

The dream that natural language can be translated by machine come true in the late twentieth. Though not completely perfect, machine translation still fulfil the requirement of translation in technical manuals, scientific documents, commercial prospectuses, administrative memoranda and medical reports.(W.John Hutchins, 1995) Researchers divide traditional machine translation method into three categories:Rule-Based, Corpus-Based and Hybrid methods, and all of them have their own merits and demerits. The first one builds the translation knowledge base on dictionaries and grammar rules, but it is not so practical for languages without much correlation and highly rely on human experience. The second one builds the translation knowledge by making full use of the corpus, which is still the mainstream of today’s machine translation. The last one mix both of rule and corpus and successfully raise the efficiency of translation, but it is tough to be managed because of complex system and weak extend ability. (Hou Qiang, 2019) According to Martin Woesle, the advantages and disadvantages of machine translation can be obvious. For advantages, machine translation has its speed and availability, low costs, efficiency and welcome of cooperation. However, it can not satisfy some special situation such as: noisy background, ill connectivity, short of electricity, corpus limitation and cultural sensitivity. (Martin Woesle, 2021)

4. Post-editing

If we try to understand post-editing literally, it can be described as “ “the correction of MT output by human translator”(Senez, 1998) or “translator use the machine translation products as the raw material to further editing and control the quality that they can satisfy special client.”[1](Zhao Tao, 2021) Generally speaking, post-editing can be divided into two types:light post-editing and full post-editing according to the level of human intervention. The former aims to produce the translation that can be almost understood and the later wants to give the production as good as human translation. But this standard is quite ambiguous. Translation Automation User Society also gave a discrimination that publication quality post-editing mainly needed in some high quality required situation, but keynote translation with high speed is more suitable for normal occasions. Despite the slight different between these two categorizations, the principle to categorize post-editing is identical: purpose.

4.1. Preparation

According to researchers, post-editing machine translation can increase the productivity of translators in terms of speed, while retaining or in some cases even improving the quality of their translations. However, such benefits are not always guaranteed except in the right condition.[2] Since the purpose of the translator is efficiency and accuracy, they have to evaluate what are right texts and what are worth to be post-edited. Firstly, it is very important to identify which kind of text should be translated by machine and worth to be post-edited. For the reason that the AI technology has been developed greatly, people always have wrong conception that machine will completely replace human being. And this kind of opinion is always so convincible. AI robots are more efficient, accurate and tolerant. For most of jobs, AI robots can perfectly finish them without expensive labor cost. But it doesn’t mean translator should give way to machine translation in any field. We have to admit that the translation quality of machine translation in general text has become considerably high and is very close to human translation satisfying the information acquisition requirement of readers. (Zhou Bin, Rao Ping, 2020). So, it is more senseful to discuss the text type that should be post-edited. TAUS also gave four situations of professional machine translation: 1. Project without abundant time 2. Project with no need for high quality 3. First version of machine translation with need for human post-editing 4. Project as a method to test errors. For the first principle, we can imagine an application situation like an international meeting. In such a context, two languages or more languages will be used and there is no time for translator to record the whole meeting and translate. The immediate reaction is pretty in need. In traditional way , the translators try to use pens, paper and marks to record the main structure of speaking and then do the translate work, which definitely challenges the translator’s ability. However, this can be change when the translators only need to check and post-edit the already exist text. The machine can record the sounds and transmit them to visible material, and then what the translator should do is to find the minor mistakes and correct them. Now, let’s come to the second principle. The readers’ purpose always leads the way translator should go. If they just want to get a rough information about a text in different language, for example, from an introduction website of production, machine translation and post-editing can absolutely do it. As for the third principle and the fourth principle, we will talk about them in sections below. In conclusion, the preparation for post-editing is so indispensable that we can’t even start our researching without describing it. It is not only related to efficiency，but also restrict the machine translation in an efficient and proper domain. In this domain, the machine translation can function well and also does waste too much spirit of the translator. It is very important to mention that the translator’s experience is not always being taken into account, and obviously novice translators are quite different from those professional translators. In this paper, we discuss the problems in a very general situation from the point view of machine translation errors for professional translators as well as student translator.

4.2. Word Errors

Considering the efficiency, we now have the first conclusion: the machine translation is able to function adequately when it is in a suitable domain, which is a critical presuppose. Then, we will try to discuss things translators should pay attention to basing on the error analysis made by other researchers in different levels. Luo Jimei in 2012 counted machine translation errors happened vehicle technology text, and found the fact that the rate of lexical errors are higher more than other kinds of errors reaching 84.13％ in the whole text. During these lexical errors, the errors of term are higher than other errors reaching 78.85%. Philip in 2017 discussed six challenges machine translation may face, and two of these challenges are related to our research today. The first one is domain mismatch and the second one is rare words, which means even the most advanced neural network machine translation can’t solve words mismatching. Cai Xinjie used C-E translation of publicity text as an example to show some types of errors machine translation may made and tried to illustrated reasons in more details . From all these studies, it is easy for us to identify some rules. And we will use these rules to analyze and explain our own ideas here. To start with, researchers have common sense that the error types of machine translation should be divided into three levels: lexical, syntactical, pragmatical. Also, it is not hard to find out the lexical level is always on the central place not only for its critical position in translation, but also for its fallibility. Finally, it is mostly the domain that become the reason these errors may made. Now, let’s talk about words which is the most fundamental element of translation and has decisive influence to the quality of translation. But it is also the most fallibly part of machine translation. The main reason for this problem is that there is always a large amount of term in an professional and special domain and machine can not recognize the context and choose the most proper meanings of a word based on the context. It is the polysemy of words that caused this problem, which can not be distinguished by the level of grammar, but lays on the level of semantic and pragmatics. Example 1 and example 2 show the application of the same word in different field. (1) Application of 3D Lidar Technology to Optimized Routine Design of Powerline A: 三维激光雷达技术在电力线常规优化设计中的应用 B: 三维激光雷达技术在输电线路优化设计中的应用

(2) Analysis on face smooth blasting A：表面光面爆破分析 B：工作面光面爆破分析

Except the translation errors in term, there are some other errors like: conjunction errors, misidentify of parts of speech, acronym errors, wrong substitute etc. Now, we will continue to talk about the second important word error—conjunction errors. Let’s see examples: (3) and alloys and compounds containing these metals A. 以及含有这些金属的合金和化合物。 B.或者含有这些金属的合金或者化合物。

(4) The others are new records or Guizhou or the mainland or China A. 其他的是新的记录或贵州或大陆或中国。 B. 其余为贵州新记录或中国大陆新记录种. From these examples, it not difficult for us to find that the translation of conjunctions ,especially when more than one conjunction, is misleading the machine and make it confused for machine to analyze which word should be in juxtaposition with another word and which word has preference relation with another word. However, what translator should focus on in post-editing is very explicit for about 78.85% errors are wrong translation of term. This part of discovery has enlightened us and helps us give some advices to post-editing translator. Firstly, when the translator tries to prepare for a post-editing job, he can try to acknowledge the type, domain or field of the text. Special dictionary, digital data may be needed to finish the work. Then, during the post-editing, the translator’s spirit, time and thought should be spent more on dealing with vocabulary and he can clearly realize that how many precents of his effort should be put on words, which greatly raise the efficiency. Finally, instead of predication that machine translation allows more and more people entering this field without strict practice and train, we would rather to believe that professionality will be more stressed on because only can a professional and skillful translator intuitively react to the term errors and erect them. Especially in some instant translation required situation, an adept translator is quite in need to solve problems. We may as well imagine a future where post-editing become increasingly a professional job and the division of the labor will be more precise and explicit.

4.3 Syntactic Errors

Newmark believes that in syntactical level, sense, signifier, coherent, and natural express response for translation. And the original language and target language should be equivalent in function. (Newmark, 1988). Based on this theory, Cai Xinjie found three syntactical errors in machine translation. The first syntactical error is logical confusion. For example: (3) Create abundant and humanistic urban space A. 创造丰富，人性的城市空间 B. 创造丰富人文的城市空间 We can find from this article that even with an and between abundant and humanistic, the machine can not divide the relationship between these two words and make a logical mistake. So this situation need the translator with a clear and logical mind. He should fully realize the logical relationship between words and words, sentences and sentences. We can now deny the idea that only the post-editing text should be read. It is still necessary for translators to scan the text and have a basic concept of the whole material, even though they may not have enough time to research the material in detail. Moreover, when we want to understand a sentence, we can’t live the help of context. There are some types of context such as context based on stories happened before, context based on situation, context based on culture. For example, we select a sentence from a story: (4) He looped the painter through a ring in his landing-stage. A. 他把油漆工从着陆台上的一个环上绕了一圈。 B. 水鼠把缆绳系在码头的缆桩上. Then we can find from this sentence that the machine can’t even give an understandable sentence without the context. That can be a very tough situation for translators because they can’t even do some minor changes according to the original machine translation text. So two strategies are considered useful for this situation. To avoid the chaos made by unrecognizable context, the translator should firstly select the right situation that machine translation can be applied. According to Tang Yefan, machine translation and post-editing can be suitable for technical text which has features of professionality, literary meaning, similar sentence model, and simple purpose. (Tang Yefan, 2018) Li Shiqi also said machine translation is efficient for mechanic text which has stable writing type and expressive methods. (Li Shiqi, 2018)But she also declare that machine translation can help her translate text with a great deal of term which a litter bit contrast to what we found, so that can be a new problem we dig deeper. Another strategy is improving the education of translators themselves. We already certified that no matter how machine translation being developed, it still can’t leave the edition of human. It is the human who translate and post-edit the text that decide the quality of translation. Errors will be made by machine, and human’s job is to realize, find and erect those errors. That is why translators should be sensible to different error types. Moreover, it is translator’s duty to know the purpose under translation. If the reader or hearer want information, then the translator give information. If the participant requires to exchange culture and reach a common view, it also the translator’s responsibility.

5. Some Other Problems

Since we discussed machine translation, post-editing and their efficiency, some researchers may long have a question: “Is machine translation post-editing worth the effort?” There are so many things have to be done before and during post-editing, and why not just pick a text and translate it? Actually some researchers have done the study about this question. Maarit Koponen in his article did a survey about post-editing and effort from point views of productivity, quality, monolingual post-editing(Maarit,). For productivity, he argues that the survey can demonstrate higher rate of productivity when the translators are doing post-editing. For quality, studies show the post-editing texts with even higher quality than manually translated texts. In the condition that readers can’t contact with original language, the correct rate of sentences will be low. As for effort, all the aspects above are only some parts of the work which can not easily take a final conclusion. And when the researchers try to interview some translators about their feeling, it can be really subjective. Everybody has his own standard to evaluate his effort in doing post-editing. To solve this problem, researchers use new technologies and methods such as: Eye tracking data, Computerized metrics, translation editing rate. All in all, Maarit answered the question” Is machine translation post-editing worth the effort?” “Yes!” Though there are so many things needed to be explore like “What is the real standard to evaluate post-editing efficiency”, ”Can machine translation be used in wider domain, especially those proved can’t be translated by machine?” or “Can post-editing finally be done by machine and human finally give way to the AI” Now, the fact we can completely say yes is that post-editing is a new and charming field for more translator to join in. From Maarit’s study, we can also give advices to translators wanting to join in this new world. Post-editing is worth doing only when translators are able to use computer software with complexity of using steps. So, it is necessary to take the pre-education and learn and practice the computer tools, which is not so easy for those who don’t have a computer or not familiar with the computer. It is a relatively narrow space for some people. Then, because of the original text is heavily influencing the result of post-editing, translators can’t just post-edit based on the machine translation raw material, which has its high requirement to translator’s reading comprehension ability and logical thinking ability. They must quickly scan the raw article or use their ears to catch information from original resource of the text. However, even though there are so many things have to be done before becoming a post-editor, translator can actually get merits from post-editing. Some dilemma in translation can be solved under the efficiency of post-editing. For translator, their career in simultaneous interpretation can be longer because the machine can help them remember, transmit and store information. All they have to do is fully understanding the original material and using their professional knowledge to post-edit the machine translated work. Simultaneous interpretation is a job with high requirement of younger people’s reaction and remembrance, that most of translator of this field have short career. But actually, it is also not so easy to become a simultaneous interpreter. Post-editing can help more interpreters to stay in the job longer. Another problem is that the salary of translator is decreasing, and at the same time the quality of translation is decreasing either. With post-editing, we may have better way to solve this problem. For example, translator need to be more professional and the quality of translation will be improved in post-editing, which in turns give more chances to translator raising their salary and expel the low-quality translator who may compete with them by lower salary requirement. For customers, post-editing makes it possible to access translation in different situation. For example, customers don’t even need to contact a translator face to face that he can enjoy the translation service everywhere and anytime with the efficiency of machine translation and post-editing. The last merit is that the development of post-editing can also prompt the development of machine translation. People will more understand how to live with machine.

6. Post-editing Application

In this part, we will find the application of post-editing in business situation. People can see that the application of post-editing is going far away from we can imagine and there is always a mature and dynamic mode around it because the user is trying to make money of it. Now, let’s find out something around this new and popular business model. To begin with, we want to introduce a concept” crowdsourcing”. Crowdsourcing means that companies entail using many disparate individuals to perform services or to generate ideas or content. (Jeff Howe, 2006) And based on this concept, a new concept rises—Online Collaborative Translation, which is a new kind translation mode covering most fields and developing rapidly. In recent year, with the cooperation of deep learning, mass data and high-performance computing, AI has great advancement. The quality of translation is rising because the neural machine translation is becoming technology mainstream. The online collaborative translation mode will not only be restricted in human-human relationship, but human-machine, machine-machine becoming possible. (Shao Lu, 2014) What is the process of online collaborative translation? There are 5 steps and every step aims to solve one or more problems: 1. Term marking 2. Term editing 3. Post-editing 4. Checking 5. Item up shelf. The first, third and fourth steps are separately taken control of one person, while the second step need to be done by all the translator. The purpose of the first step is to mark some special terms and categorize them so that the post-editor can easily recognize those special terms. The second step aims to check and edit term that them can be reused in other articles with the same style or topic. The third step aims to erect grammar errors and semantic errors that make the article influent and readable. The last step will need to be done by professional and skilled translators that problems not so obvious can be found. Then we will explain this mode with a real practice example. However, what most surprise us is that this kind of mode is mostly applied in the translation of online novels. But it doesn’t mean that our finding of post-editing is wrong because the online novels are so special that only their plots are important to readers, and expression, text and style are not so important, because they are only serving for plots, which exact follow the principle post-editing obey. On the working website, we can see the original passage is lying on the left side and the term base is on the right side which can help translator to search terms easily. The translator can also add new terms or correct old terms if they are not satisfied with exist terms. Then the original passage will be post-edited one sentence to another.

Original sentence：“小样，有本事你就把小爷我给劈了！”季风烟躲开一道天雷的瞬间，朝着天空比了一个嚣张至极的中指。 Machine translation version: ”Little brat, if you have the ability, you’ll chop me up! The moment Ji Fengyan dodged a bolt of lightning, he pointed at the sky with an extremely arrogant middle finger.

Step one: “小样，有本事你就把小爷我给劈了！”季风烟躲开一道天雷的瞬间，朝着天空比了一个嚣张至极的中指。 “Little brat, if you have the ability, you’ll chop me up! “The moment Ji Fengyan dodged a bolt of lightning, he pointed at the sky with an extremely arrogant middle finger. (Highlighting terms)

Step two: “小样，有本事你就把小爷我给劈了！”季风烟躲开一道天雷的瞬间，朝着天空比了一个嚣张至极的中指。 “Small kind of, you would chop me up if you had power! After dodging a bolt of lightning, Ji Fengyan pointed at the sky with an middle finger showing the arrogant attitude of him.”

Step three: Translators read the last chapter and the next chapter that they can understand the context.

From this example, we find out that terms are still the first and the most important problem should be solved, and in practice, people find ways to overcome it and make the quality of translation better. But except post-editing, we can see here a pre-editing step which is a good complementary way to post-editing and make it possible the collaborative translation online. Let’s give another example.

Original sentence：雪宝赌气，扔给她一句：“我死了也是你害的！” Machine translation version: Carol Faiman was angry and threw her a sentence,” You killed me too!”

Step one: 雪宝赌气，扔给她一句：“我死了也是你害的！” Carol Faiman was angry and threw her a sentence,” You killed me too!” (Highlighting terms)

Step two: 雪宝赌气，扔给她一句：“我死了也是你害的！” Carol Faiman was angry and threw her a words,” You killed me too!”

With these practical examples, we now drilled deeper in post-editing.

Conclusion

Believe it or not, machine translation will move from a periphery place to central place. The technology is developing and everything changes day and night. What we should do is to identify again and again our human’s position. Machine is just a tool and only human can make good use of this tool. In the passage, we firstly discussed functional equivalence and skopos theory which are important principles translator should obey in doing their works. We try to do the research in three levels: lexical, syntactical and style. Every level has its own points. For the first level—word, which is the most fundamental part of translation, translators should be aware of the term error and the conjunction error because they occupied the most space of the lexical error. This discovery reminds us that professional and susceptive translator will be more suitable to take charge of post-editing work. Then we came to the syntactic part. In this part, logical sentence order and context are major points we talked about. In section 6, the online collaborative translation gives us inspiration: Since the translation can leave original passages, why not support the translator with context and make it convenient for them to search an check the context. Then we discussed the efficiency of post- editing, and evidence shows that post-editing is more efficient than pure human translation. In the last section, we expound a real application of post-editing and detect that post-editing has been deeply used in business practice. However, there are still some trouble waiting for researchers to find out the answer: Can post-editing be applied in broader places? Can some basic grammar errors disappear in machine translation? Can machine do the post-editing work? Some obstacles can be surmount with the development of technology.

References

Hutchins W J. Machine translation: A brief history[M]//Concise history of the language sciences. Pergamon, 1995: 431-445. Koehn P, Knowles R. Six challenges for neural machine translation[J]. arXiv preprint arXiv:1706.03872, 2017. Koponen M. Is machine translation post-editing worth the effort? A survey of research into post-editing and effort[J]. The Journal of Specialised Translation, 2016, 25: 131-148. Moratto, Riccardo, and Martin Woesler, eds. Diverse Voices in Chinese Translation and Interpreting: Theory and Practice. Springer Nature, 2021. Newmark P. A textbook of translation[M]. New York: Prentice Hall, 1988. Nord C. Translating as a purposeful activity: Functionalist approaches explained[M]. Routledge, 2014. Senez D. Post-editing service for machine translation users at the European Commission[J]. Translating and the Computer, 1998, 20. Howe J. The rise of crowdsourcing[J]. Wired magazine, 2006, 14(6): 1-4. 蔡欣洁,文炳. 汉译英机器翻译错误类型统计分析——以外宣文本汉译英为例[J]. 浙江理工大学学报(社会科学版), 2021, 46(2): 162-169. 郭建中. 当代美国翻译理论[M]. 湖北教育出版社, 2000. 侯强, 侯瑞丽. 机器翻译方法研究与发展综述 2019年3月12日[J]. 计算机工程与应用, 2019. 李诗琪. "机器翻译+译后编辑"模式在法律翻译中的应用[D]. 上海外国语大学. 罗季美, 李梅. 机器翻译译文错误分析[J]. 中国翻译, 2012, 33(5):6. 唐叶凡. 机器翻译+译后编辑在不同类型文本中的适用性分析[D]. 上海外国语大学. 王华树, 王鑫. 人工智能时代的翻译技术研究:应用场景,现存问题与趋势展望[J]. 外国语文, 2021, 37(1):9. 赵涛. 机器翻译译后编辑的现状与问题[J]. 外语教学, 2021, 42(4):5. 周斌, 饶萍. 基于实例的机器翻译评测及译后编辑修正模式[J]. 浙江理工大学学报：社会科学版, 2020, 44(3):9.