revise my essay

Introduction

With the development of big data, cloud computing and artificial intelligence, many technological innovations have emerged, and people's life has become more and more convenient. Machine translation is one of the most important technologies. It refers to the automation technology that can translate oral or written text from one language to another without human participation. As the Internet has opened up a wider multilingual world for people, this language service has become very valuable.In the past few years, machine translation research and development is amazing. Back in 2016, Google Translate launched neural machine translation, and it uses phrase based machine translation in order to reduce the gap between human translation and machine translation. This review will focus on the two question based on the development of Google Translate: 1)how does Google Translate work? 2)would it replaces the human translation?

Machine Translation: Google Translate

In terms of the nature of machine translation, it is a translation method that will automatically rely on computer technology and information technology, also the translation principles play an important role. The methods of machine translation include the literal translation method, the conversion method achieved through cohesion and conversion, and the intermediate language method that achieves the purpose of translation through feedback from the intermediate language. The principle of these three methods is to analyze and study the language in order to achieve the purpose of the translation.Different translation methods have different levels of interpretation.

As for literal translation, it is a kind of translation method which is more direct and does not need to analyze and study the language, which is the most common method for machine translation. It belongs to a kind of translation method based on the level of comparison in translation work. The latter two methods need to analyze and study the language to a certain extent and obtain the translation results on this basis.

Neural Machine Translation(NMT ）

According to the sure-language website, it says Google Translate is mainly based on neural machine translation (NMT).

NMT uses neural network-based technology to achieve more context-accurate translation instead of translating broken sentences one word at a time. Using a large artificial neural network to calculate the probability of a word sequence, NMT puts the complete sentence into an integrated model.

NMT can learn and collect information, aiming to imitate the neurons of the human brain, establish connections, and evaluate input as a whole unit. NMT is analyzed in two stages: encoding and decoding. In the encoding stage, text from the source language is input into the machine and then classified into language vectors. Words that are similar in the context will be placed in a comparable word vector. Next, the decoding stage effectively and seamlessly sends the vector to the target language. Throughout the translation process, technology is more than just translating words and phrases; instead, it is translating context and information.

Figure 1.NMT or SMT: Case Study of a Narrow-domain English-Latvian Post-editing Project

However, the NMT cannot translate 100% accurate either. In the Case Study of SMT and NMT, it shows the NMT even has a higher chance to receive the mistranslation error(Pinnis & Skadiņa, 2017). Which decreases the accuracy of machine translation.

Challenges of Machine Translation

Machine translation is fast, convenient, and has high cost-effectiveness. It can translate large amounts of text in seconds. At the same time, machine translation(for non-informal translate) is free, and there are hundreds of applications that can translate text, images even voice anytime and anywhere with the press of a finger. Not only that, a machine translator can translates hundreds of different languages. This are the highlights of machine translation, which can not achieve by human.

But after this, there are still many weakness in machine translation.

First of all, machine translation is not culturally sensitive. Which means machine cannotg understand or recognize slang, jargon, puns and idioms. There is an example of Google Translate in

Figure 2. Google Translate https://translate.google.com/?ui=tob&sl=auto&tl=en&text=%E4%BA%BA%E5%B1%B1%E4%BA%BA%E6%B5%B7&op=translate

The phrase 人来人往(ren lai ren wang) means prosperous in Chinese, even Google Translate got the literal meaning correct, but it’s hard to interpret the intention based on “People come and go”.

Different cultures have different language systems, and it’s hard for human to be able to program machines to understand or experience a particular culture. Therefore, the translation produced may not conform to the cultural values and specific norms. This is one of the challenges that machines need to overcome.

Secondly, machine translation can not connect words in context. In many languages, the same word may have multiple completely unrelated meanings. In this case, context will have a great impact on the meaning of words, and the understanding of word meaning depends on the context, which is an endless cycle. With current technology, only human beings can combine words with context by determine their actual meaning in sentence. Also, only human beings can translate smoothly in the use of diction. For machine translation, this is undoubtedly very difficult.

Conclusion

Machine translation can replace people to do part of the work in large-scale, fast, and high-quality scenarios. However, it is impossible to achieve high-quality. Fully automatic machine translation can only use in a particularly narrow field. It is unlikely that machine translation will replace humans, at least in the foreseeable future. Especially in some professional fields, it is almost impossible for the current quality of translators to exceed that of high-level translators. Although the effectiveness of human translation is not as good as machine translation, machine translation is not as accurate as human translation. Putting aside the shortcomings of these two, accuracy is the most important thing in translation. If you can't express the meaning correctly, how long it takes is useless. These two kinds of translation services have their own advantages.

However, we can see that most human translation services now use certain auxiliary translation software.Beside their own advantages, combining the two translations does a win-win strategy. The combination of human and machine not only saves time and guarantees quality, but also conforms to the future development trend.

References

Pinnis, M., & Skadiņa, I. (2017). NMT or SMT: Case Study of a Narrow-domain English-Latvian Post-editing Project [Review].

How does Google Translate work and is it any good? (2019, November 12). Retrieved November 10, 2020, from https://www.sure-languages.com/how-does-google-translate-work-and-is-it-any-good/

Costa-jussa, Marta & Rapp, Reinhard & Lambert, Patrik & Eberle, Kurt & Banchs, Rafael & Babych, Bogdan. (2016). Hybrid Approaches to Machine Translation. 10.1007/978-3-319-21311-8.