Machine Trans EN 10
Chapter 10 熊敏 Research on the English Chinese Translation Ability of Machine Translation for Various Types of Texts
机器翻译对各类型文本的英汉翻译能力探究
熊敏, Xiong Min, Hunan Normal University
Abstract
The history of machine translation can be traced to 1940s, undergoing germination, recession, revival and flourish. Nowadays, with the rapid development of information technology, machine translation technology emerged and is gradually becoming mature. Whether manual translation can be replaced by machine translation becomes a widely-disputed topic. So in order to explore the ability of machine translation, I adopts two versions of translation, which are manual translation and machine translation(this paper uses Youdao translation) for different types of texts(according to Peter Newmark's types of text：informative text, expressive text and vocative text). The results are quite different in terms of quality and accuracy. The results show that machine translation is more suitable for informative text for its brevity, while the expressive texts especially literary works and vocative texts are difficult to be translated, because there are too many loaded words and cultural images in expressive text and dependence on the readers. To draw a conclusion, the relationship between machine translation and manual translation should be complementary rather than competitive.
Key words
machine translation; manual translation; Newmark's type of texts
题目
Research on the English Chinese Translation Ability of Machine Translation for Various Types of Texts
摘要
机器翻译的历史可以追溯到上个世纪40年代，经历了萌芽期、萧条期、复兴期和繁荣期四个阶段。如今，随着信息技术的高速发展，机器翻译技术日益成熟，于是机器翻译能不能替代人工翻译这个话题引起了广泛热议。为了探究机器翻译的能力水平是否足以替代人工翻译，本人根据皮特纽马克的文本类型分类理论（信息类文本，表达文本，呼吁文本），选择了相应的译文类型，并且将其机器翻译的版本以及人工翻译的版本进行对比。结果发现，就质量和准确度而言，译文的水平大相径庭。因为其简洁的特性，机器比较适合翻译信息文本。而就表达文本，尤其是文学作品，因其存在文化负载词及相应的文化现象，所以机器翻译这类文本存在难度。而呼唤文本因其目的性以及对读者的依赖性较强，翻译难度较大。由此得知，机器翻译和人工翻译的关系应该是互补的，而不是竞争的。
关键词
机器翻译；人工翻译；纽马克文本类型
1. Introduction
1.1.Introduction to machine translation
Machine translation, also known as computer translation is a technique to translating through machine translator, such as Google translation and Youdao translation. Machine translation is one of the branches of computational linguistics, ranging from computer science, statistics, information science and so on. Machine translation plays an important role in all aspects. Machine translation can be traced back to 1940s, when British engineer Booth and American engineer Weaver proposed using computer to translate and started to study machines used for translation. And the first machine translating system launched in 1954, used in English and Russian translation. And this means machine translation becomes a reality. However, the arrival of new things always accompanies barriers and setbacks. In the 1960s, reports from ALPAC (Automated Language Processing Advisory Committee) showed studies on machine translation had stagnated for a decade. At that time, due to the high cost of machine, choosing human translators to finish translating works was more economical for most companies. What’s worse, machine had difficulty in understanding semantic and pragmatic meanings. Fortunately, in the 1970s, with the advance and popularity of computer, machine translation was gradually back on track. With the development of science and technology, machine translation has better technological guarantee, such as artificial intelligence and computer technology. In the last decades, from the late 90s till now, machine translation is becoming more and more mature. The initial rule-oriented machine translation has been updated and Corpus-aided machine translation appears. And nowadays, technology in voice recognition has been further developed. This technique is frequently applied in speech translation products. Users only need to speak source language to the online translator and instantly it will speak out the target version. But machine can’t fully provide precise and accurate translation without any grammatical or semantic problems. Machine can understand the literal meaning of texts, but not the meaning in context. For example, there is polysemy in English, which refers to words having multiple meanings. So when translating these words, translators should take contexts into consideration. But machine has troubles in this way. In addition, machine has no feeling or intention. Hence, it is also difficult to convey the feelings from the source language.(Wei 2021:5)#
1.2.Process of machine translation
The process of manual translation is different from that of machine translation. Here is the process of the former. (1) Understand source language. (2) Use target language to organize language. (3) Generate translation. Unlike manual translation, machine translation tends to analyze and code source language first, then look for related codes in corpus, and work out the code that represents target language, generating translation. But they share a common feature, which is that Lexicon, grammatical rules and syntactic structure are taken into consideration. This is one of the biggest challenges for machine translation.
2.Theorectical framework
2.1Newmark’s text typology
Text typology is a theory from linguistics. It undergoes several phrases. Karl Buhler, a German psychologist and linguist defines communicative functions into expressive, information and vocative. Then Roman Jakobson proposes categorization of texts, which are intralingual translation, interlingual translation and intersemiotic translation. Accordingly, he defines six main functions of language, including the referential function, conative function, emotive function, poetic function, metalingual function and the phatic function. Based on the two theories, Peter Newmark, a translation theorist, summarizes the language function into six parts: the informative function, the vocative function, the expressive function, the phatic function, the aesthetic function, and the metalingual function. Later, he further divided texts into informative type, expressive text and vocative type according to the linguistic functions of various texts. (Newmark 2002:2)# The core of the informative texts is the truth. It is to convey facts, information, knowledge of certain topics, reality and theories. The language style of the text is objective, logical and standardized. Reports, papers, scientific and technological textbooks are all attributed to informative texts. Translators are required to take format into account for different styles. The core of the expressive text is the sender’s emotions and attitudes. It is to express sender’s preferences, feelings, views and so on. The language style of it is subjective. Imaginative literature, including fictions, poems and dramas, autobiography and authoritative statements belong to expressive text. It requires translators to be faithful to the source language and maintain the style. The core of the vocative text is readership. It is to call upon readers to act, think, feel and react in the way intended by the text. So it is reader-oriented. Such texts as advertisement, propaganda and notices are of vocative text. Newmark noted that translators need to consider cultural background of the source language and the semantic and pragmatic effect of target language. If readers can comprehend the meaning at once and do as the text says, then the goal of information communication is achieved. (Liu 2021:3)# To draw a conclusion, informative texts focus on the information and facts. Expressive texts center on the mind of addressers, including feelings, prejudices and so on. And vocative texts are oriented on readers and call on them to practice as the texts say.
2.2Study method
Manual translations of the three texts are selected from authoritative versions and universally acknowledged. And machine translations of those come from Youdao Translator. And in this thesis I will compare and evaluate the two methods in word diction, sentence structure, word order and redundancy.
3.Comparison and analysis of machine translation and manual translation
3.1Informative text
（1）English into Chinese
①Source language:
Keep the tip of Apple Pencil clean, as dirt and other small particles may cause excessive wear to the tip or damage the screen of i-pad.
Target language:
Machine translation: Apple Pencil笔尖应保持清洁，灰尘等小颗粒可能会导致笔尖过度磨损或损坏ipad屏幕。
Manual translation: 保持Apple Pencil铅笔的笔尖干净，因为灰尘和其他微粒可能会导致笔尖的过度磨损或损坏iPad屏幕。
Analysis: Here is the instruction of Apple Pencil. And the manual translation is the Chinese version on the instruction.Product instruction tends to be professional, since there are many terms for some concepts. Machine can easily identify these terms and provide related words to translate. The machine version is faithful and expressive to the source language. So it is well-qualified and readable for readers to understand the instruction. So we can use machine to translate informative text.
②Source language:
China on Saturday launched a rocket carrying three astronauts-two men and one woman - to the core module of a future space station where they will live and work for six months, the longest orbit for Chinese astronauts.
Target language:
Machine translation: 周六，中国发射了一枚运载三名宇航员(两男一女)的火箭，进入未来空间站的核心舱，他们将在那里生活和工作6个月，这是中国宇航员最长的轨道。
Manual translation: 周六，中国发射了一枚搭载三名宇航员(两男一女)的火箭，进入未来空间站的核心舱，他们将在那里生活和工作6个月，这是中国宇航员最漫长的一次轨道飞行。
Analysis: This is a news from Reuters, reporting that China has launched a rocket.The meaning of the two translations is almost the same, except for some word diction. But there are some details dealt with different choice. For example, the last sentence of the machine translation is a bit of obscure and direct. There are some ambiguous words and expressions.
(2)Chinese into English
Source language:湖南省博物馆是湖南省最大的历史艺术类博物馆，占地面积4.9万平方米，总建筑面积为9.1万平方米，是首批国家一级博物馆，中央地方共建的八个国家级重点博物馆之一、全国文化系统先进集体、文化强省建设有突出贡献先进集体。
Target language: Manual translation: As the largest history and art museum in Hunan province, the Hunan Museum covers an area of 49,000㎡, with the building area reaching 91,000㎡. It is one of the first batch of national first-level museums and one of the first eight national museums co-funded by central and local governments.
Machine translation: Museum in hunan province is one of the largest historical art museum in hunan province, covers an area of 49000 square meters, a total construction area of 91000 square meters, is the first national museum, the central place to build one of the eight national key museum, national cultural system advanced collectives, strong culture began with outstanding contribution of advanced collective.
Analysis: Machine translation is not faithful enough in content. For instance, “首批国家一级博物馆” is translated into “first national museum”, which is not the meaning of the source language. And there are some obvious grammar mistakes in the machine translation. For example, machine translates it into just one sentence but there are multiple predicates in it. So it is not grammatically permissible. What’s more, the sentence structure of machine translation is confusing and the focus is not specific enough.
3.2Expressive text
(1)English into Chinese
Source language:
An individual human existence should be like a river- small at first, narrowly contained within its banks, and rushing passionately past rocks and over waterfalls. Gradually the river grows wider, the banks recede, the waters flow more quietly, and in the end, without any visible breaks, they become merged in the sea, and painlessly lose their individual being.()
Target language:
Machine translation: 一个人的存在应该像一条河流——开始很小，被紧紧地夹在两岸中间，然后热情奔放地冲过岩石，飞下瀑布。渐渐地，河面变宽，两岸后退，水流更加平缓，最后，没有任何明显的停顿，它们汇入大海，毫无痛苦地失去了自己的存在。
Manual translation:人生在世，如若河流；河口初始狭窄，河岸虬曲，而后狂涛击石，飞泻成瀑。河道渐趋开阔，峡岸退去，水流潺缓，终了，一马平川，汇于大海，消逝无影。
Analysis: Here is a well-known metaphor in the prose How to Grow Old written by Bertrand Russell. The manual translation is written by Tian Rongchang.This is a philosophical prose with graceful language. Literary translation is a most important and difficult branch of translation. Translator should focus on the literal meaning, culture, writing style and so on. It is a combination of beauty and elegance. Therefore, translators find it in a dilemma of beauty and faithfulness, let alone translating machine. Compared with manual translation, machine translation has difficulty in word choice. It is faithful and expressive, but not elegant enough.
(2)Chinese into English
Source language:没有一个人将小草叫做“大力士”，但是它的力量之大，的确是世界无比。这种力，是一般人看不见的生命力，只要生命存在，这种力就要显现，上面的石块，丝毫不足以阻挡。因为它是一种“长期抗战”的力，有弹性，能屈能伸的力，有韧性，不达目的不止的力。(Zhang, 2007:186)#
Target language:
Machine translation: No one calls the little grass "hercules", but its power is truly matchless in the world. This force is invisible life force. As long as there is life, this force will show itself. The stone above is not strong enough to stop it. Because it is a "long-term resistance" of the force, elastic, can bend and extend force, tenacity, not to achieve the purpose of the force.
Manual translation: Though nobody describes the little grass as a “husky”, yet its herculean strength is unrivalled. It is the force of life invisible to naked eye. It will display itself so long as there is life. The rock is utterly helpless before this force- a force that will forever remain militant, a force that is resilient and can take temporary setbacks calmly, a force that is tenacity itself and will never give up until the goal is reached. (by Zhang Peiji)
Analysis:This is the excerpt of a well-known Chinese prose written by Xia Yan. It is written during the war of Resistance Against Japan. So the prose holds symbolic meaning, eulogizing the invisible tenacious vitality so as to encourage Chinese to have confidence in the anti-aggression war. Compared with manual translation, machine translation is much more abstract and confusing, especially for the word diction. For example, “大力士” is translated into “hercules” which is a man of exceptional strength and size in Greek and Roman Mythology, making it difficult to understand if readers of target language have no idea of the allusion. What’s worse, the machine version doesn’t reveal the symbolic meaning of the text, which is the core of this prose.
3.3Vocative text
(1)English into Chinese
①Source language:
iPhone went to film school, so you don’t have to. (Advertisement of iPhone13)
Target language:
Machine translation: iPhone上的是电影学院，所以你不用去。
Manual translation:电影专业课，iPhone同学替你上完了。
Analysis：Here are advertisements of iPhone on Apple official website. There is a personification in the source language. It is used to stress the advancement and proficiency in camera, which is an appealing selling point to potential buyers. Compared with manual translation, machine translation is plain and not eye-catching enough for customers.
②Source language:
5G speed OMGGGGG
Machine language: 5克的速度 OMGGGGG
Manual translation:
iPhone的5G 巨巨巨巨巨5G
Analysis: The “G” in the source language is the unit of speed, standing for generation. However, it is mistaken as a unit of weight, representing gram in the machine translation. So the meaning is not faithful to the source language at all. As for manual translation, it complies with the source in form. Specifically speaking, five “G”s in the former complies with five characters “巨”in the latter. And the pronunciation of the two is similar. There are two layers of meaning for the 5 “G”s. One exclaims the fast speed of 5 generation network and the other new technology. In the manual version, “巨”can be used to show degree, meaning “quite” or “very”.
③Source language:
History, faith and reason show the way, the way of unity. We can see each other not as adversaries but as neighbors. We can treat each other with dignity and respect, we can join forces, stop the shouting and lower the temperature. For without unity, there is no peace, only bitterness and fury."
Target language:
Machine translation: 历史、信仰和理性指明了团结的道路。我们可以把彼此视为邻居，而不是对手。我们可以尊严地对待彼此，我们可以联合起来，停止大喊大叫，降低温度。因为没有团结，就没有和平，只有痛苦和愤怒。
Manual translation:历史、信仰和理性为我们指明道路。那是团结之路。我们可以把彼此视为邻居，而不是对手。我们可以有尊严地相互尊重。我们可以联合起来，停止喊叫，减少愤怒。因为没有团结就没有和平，只有痛苦和愤怒
Analysis: Speech is a way to propagate some activity in public. It is an art to inspire emotion of the audience. The source language is the excerpt of Joe Biden’s inaugural speech. The speech should be inspiring and logic. The machine translation has some misunderstanding. Taking the translation of “lower the temperature” for example, machine only translates its literal meaning, relating to the temperature itself, without considering the context. What’s more, it is less logic than the manual one. Therefore, it adds difficulty to inspire the audience and infect their emotion.
4.Common mistakes in machine translation
4.1 lexical mistakes
Common lexical mistakes include misunderstandings in word category, lexical meaning and emotive and evaluative meaning. Misunderstanding in word category shows in the classification of word in the source language. As for misunderstanding in lexical meaning, machine has difficulty in precisely reflecting the meaning of the original texts, due to different cultural background and different language system. And for misunderstanding in emotive meaning, machine has no intention and emotion like human-beings. Therefore, it’s impossible for it to know writers’ feelings and their writing purposes. So sometimes, it may translate something negative into something positive. (Wang 2008:45)#
4.2 grammatical mistakes
Grammatical analysis plays an important part in translation. Normally speaking, every language has its own unique grammatical rules. So in the process of translation, if translators don’t know the formation rule well, the sentence meaning will be affected. Even though all the lexical meanings are well-known by translators, the lack of consciousness of grammaticality makes it harder to arrange words according to sequential rule. English tends to be hypotactic, while Chinese tends to be paratactic. English sentences are connected through syntactic devices and lexical devices. While Chinese sentences are semantically connected, which means there are limited logical words and connection words in Chinese. So when translating English sentence, we should first analyze its grammaticality and logical structure and then rearrange its sequence. However, online translating machine has troubles in grammatical analysis, which makes its improvement more difficult.
4.3 other mistakes
The two mistakes above are the internal ones. Apart from mistakes in linguistic system, there are some mistakes in other aspects, such as cultural background.
5.Reasons for its common mistakes
5.1 Difference in two linguistic system
With different history, English and Chinese have different ways of expression. Commonly speaking, English is synthetic language which expresses grammatical meaning through inflection such as tense and Chinese is analytic language which expresses grammatical meaning through word order and function word. In addition, English is more compact with full sentences. Subordinate sentence is one of the most important features in modern English. Chinese, on the other hand, is more diffusive with minor sentences.
5.2 Difference in thinking patterns and cultural background
According to Sapir-Whorf’s Hypothesis, our language helps mould our way of thinking and consequently, different languages may probably express their unique ways of understanding the world. For two different speech communities, the greater their structural differentiations are, the more diverse their conceptualization of the world will be. For example, western culture is more direct and eastern culture more euphemistic. What’s more, English culture tends to be individualism, focusing on detail, through which it reflects the whole, while Chinese culture tends to be collective. Different thinking patterns will add difficulty for machine to translate texts.
5.3 Limitation of computer
Recently, there are some breakthroughs and innovation in machine translation. However, due to its own limitation, online translation has limitation in some ways. Firstly, compared with machine, human brain is much more complicated, consisting of ten billions of neuron, each of which has different function to affect human’s daily activities and help humans avoid some errors. However, computer can only function according to preset programming has no intention or consciousness. Until now, countless related scholars have invested much time in machine translation. They upload massive language database, which include almost all linguistic rules. But computers still fail to precisely reflect the meaning of source language for many times due to the complexity and flexibility of language. On the other hand, computers can’t take context into consideration. During translation, it is often the case that machine chooses the most-frequently used meaning of one word. So without the correct and exact meaning, readers are easier to feel confused and even misunderstand the meaning of source language. (Qiu 2021:4)#
6.Conclusion
From the analysis above, we can draw a conclusion that machine deals with informative text best, followed by non-literary translation of expressive text. What’s more, machine can be a useful tool to get to know the gist and main idea of a specific topic, for the simple sentence structure and numerous terms. And it can improve translating efficiency with high speed. But machine has difficulty in translating literary works, especially proses and poems.
Machine translation has mixed future. From the perspective of commercial, machine translation boasts a bright future. With the process of globalization, the demand for translation is increasing accordingly. On one hand, if we only depend on human translator to deal with translating works, the quality and accuracy of translation can be greatly affected. On the other hand, if machine is used properly to do some basic work, human translators only need to make preparation before translating, progress, polish and other advanced work, contributing to highly-qualified translation and high working efficiency.
However, compared with manual translation, machine translation has a bleak future. It is still impossible for machine to replace interpreter or translator in a short term. With intelligence and initiative, humans are able to learn new knowledge constantly, which machine will never accomplish. Besides, machine is not used to replace translators but to assist them in work. In other words, translators and machine carry out their own duties and they are not incompatible.(He 2021:5)#
To draw a conclusion, although there are certain limitations of machine translation, it can serve as a catalyst for translating works. Therefore, with the rapid development of artificial intelligence and related technology, there are still many opportunities for machine translation.
In my opinion, it would be better if the first letter of the subtitles are being capitalized. Corrected by--Chen Huini (talk) 05:13, 15 December 2021 (UTC)Chen Huini
Reference
Chen Cheng陈诚.机器翻译技术的综述[J][Overview of Machine Translation Technology].Electronic Techonology 电子技术,2021,50(11):290-291.
Cui Zihan 崔子涵.机器翻译译文质量对比——以谷歌翻译和DeepL为例[J] [Comparison among Machine Translation--Taking Google Translation and Deepl for Example].Overseas English 海外英语,2021(15):182-183.
He Xinyu何馨宇.机器翻译的发展及其对翻译职业化的影响研究[J] [The Development of Machine Translation and its Effect on Professional Transltors].Overseas English 海外英语,2021(20):48-49.
He Wen 何雯, Wang Xiufeng 王秀峰.信息型文本的在线机器翻译错误研究[J][Research on Errors in Online Machine Translation of Informative text ].Overseas English海外英语,2021(15):188-189.
Li Deyi 李德毅. (2018). 人工智能导论 [Introduction to Artificial Intelligence]. Beijing: China Science and Technology Press 中国科学技术出版社.
Liu Qin刘琴.功能目的论对于不同文本类型的翻译解读[J][Analysis of Translations in Different Types of Text based on Functionalist Approaches].Overseas Engliosh 海外英语,2021(17):8-9.
Li Hanji 李晗佶. (2021). 人工智能时代翻译技术与译者关系演变与重构 [Evolution and reconstruction of the relationship between translation technology and translators in the era of artificial intelligence]. 西华师范大学学报(哲学社会科学版) Journal of West China Normal University (PHILOSOPHY AND SOCIAL SCIENCES EDITION) (2021-12-04) 1-6.
(英) Peter Newmark A Textbook of Translation[M] Shanghai Foreign Education Press, 2002
Qiu Quanju 仇全菊.大数据时代背景下机器翻译及其发展趋势[J][Machine Translation and its Development Trend under the Background of Big Data Era]. English Teachers 英语教师,2021,21(16):60-62.
Wang Hongqi 王红旗. (2008). 语言学概论 [Introduction to Linguistics]. Beijing: Peking University Press 北京大学出版社.
Wei Guang魏光. 人工翻译与机器翻译译文编辑比较研究[J][Comparative Study of Translation Editing between Manual Translation and Machine Translation]. Overseas English 海外英语,2021(19):18-19+21.
Zhuo Jianbin 卓键滨,Liu Wenxian 刘文娴,Peng Zili 彭子莉.机器翻译对各类型文本的德汉翻译能力探究[J][Research on the German Chinese Translation Ability of Machine Translation for Various Types of Texts]. Comparative Study of Cultural innovation 文化创新比较研究,2021,5(28):122-125.
Zhang Peiji 张培基.英译中国现代散文选[M][Selected Modern Chinese Prose Writings]. Shanghai Foreign Languages Education Press 上海外语教育出版社, 2002.