Human Translation vs Machine Translation

Back in the 1970s, our eyes widened every time Captain Kirk communicated on television from the Enterprise with inhabitants of other planets through a simple device. While it is true that this is not Star Trek and that there is still a long way to go for the universal translator to exist, The translation world has come a long way since then.

In this globalized society, translation is becoming more and more important. And it is not for less: it allows to eliminate communication barriers between totally different cultures. However, it can be a bit difficult for translators to find opportunities at a time as important as this if you need a good translation service in UAE Al Wadi Translation is the best choice. Automation is completely taking over your industry, even though we are a long way from achieving a high-quality fully automatic system.

Decades ago, the translation of large volumes of words was done by a huge team of specialists who worked on it for weeks. Obviously, the cost was prohibitive. This is changing thanks to technology. The automatic translation enables companies to achieve greater productivity at a lower cost. But how does this technology work?

Automating language

The idea of ​​automating language is quite old, although it did not become possible until the existence of computers. Decades ago, this automation was carried out exclusively through an approach based on linguistic knowledge. A committee of experts made a great effort to introduce into the machine an enormous amount of morphological, syntactic, and semantic information of each of the languages ​​in order to be able to correlate them with each other. This was such a big job that it was only interested in doing it for the most common languages, such as English or Spanish.

A few years later statistical methods were introduced. In them, what is tried is to achieve a function that estimates the probability that a sentence of a given input language corresponds to another sentence of the output language. This requires large volumes of bilingual text to enable the machine to correctly perform language processing.

There are two examples of well-known translators: Google Translate and Bing Translator. Google translator has been active since 2006. Initially, it was launched as a language-based machine translation tool, but the following year it was relaunched by adapting the statistical method. On the other hand, Bing Translator, which started working in 2009, was created by Microsoft to be a statistical machine translation machine.

Redefining the role of linguists

Despite the fact that it is now so simple and cheap to get a decent translation, we must be realistic and understand that machine translation is not for now (and probably will not be for a long time) capable of reproducing certain linguistic nuances characteristic of human beings. This is where the existence of a translator in Dubai is necessary to correct and improve the work carried out by the machine. There are many sectors in which this function is essential, such as in editorial or marketing texts, in which a cultural adaptation is even necessary that can only be carried out by the expert hands of a linguist.

It also happens that, as the volume of translated content increases, there is much more demand from the entire world population to receive new content translated into their mother tongue. This creates a wealth of opportunities for linguists to assert themselves and show that they make a difference. They may charge a little less, but they will certainly still be very valuable in this new market.

Neural Machine Translation (NMT) is the new approach to machine translation. NMT works with an end-to-end architecture that aims to train all components simultaneously to maximize their performance. The architecture takes the entire sentence into account as context, allowing you to achieve a fluent translation.

 

Has machine neural translation really caught up with human translation?

Recently, Google, Microsoft, and SDL have argued that neural machine translation (NMT) has been equated with human translation in" Google's Neural Machine Translation System: Bridging the gap between human and machine translation", "Achieving human parity on automatic Chinese to English news translation" and "SDL cracks Russian-to-English translation", respectively.

 

In a recent work belonging to the EMNLP 2018 conference, experiments are being carried out comparing neural machine translations with human translations. The task consists of classifying 55 documents and 120 sentences from the WMT 2017 Chinese-English test set. Documents and sentences are evaluated under monolingual (only text in the target language) and bilingual (text in the source language and in the target language). The evaluators are professional translators with at least three years of experience who have received positive reviews from clients. For the monolingual condition, they recruited 5 native English translators, while for the bilingual condition, they recruited 2 native Chinese translators, 1 native English translator, and 1 native translator in both English and Chinese.

 

In the monolingual condition, translators preferred man-made text to machine-produced text, both in terms of sentences and documents. In the bilingual condition, the ratings of the translators showed a significant preference for human translation over machine translation when evaluating documents. However, they did not show preference when evaluating isolated sentences, since machine translation achieves parity with the human being.

 

This is undoubtedly a good find. The quality of the NMT is impressive, but there are two important things to consider. The first is that the authors are cautious in concluding that the results could lead us to believe that machine translation (MT) performs better on adequacy than fluency. However, the TA assessment may probably be more favorable when the majority of the translators are native speakers of the source language. The second is that sentence-level assessment may be insufficient as textual, cultural, and other contexts are unknown, and these elements must be taken into account to truly understand the translation.

 

These results confirm the need to continue researching at the documentary level as in recent works. By increasing context at the document level, machine translation can improve the consistency and cohesion of the translated text. Document-level NMT can avoid some errors that are impossible to recognize at the sentence level, such as gender concordance across sentences.


Comments

Post a Comment

Popular posts from this blog

Complete Guide On The Women's Blouse | Choose New Fashion In This Winter

Image
Buy Now The women's blouse has become a staple of the female wardrobe. Originally reserved for men, the blouse is now worn by many women, businesswomen or not. The women's shirt differs from the blouse by its much more rigid appearance. Indeed, the women's shirt is a feminine variant of the men's shirt, so it is more formal than a blouse. For example, at the collar and wrists, a woman's shirt will be stiffer than a blouse. The fabrics used for a shirt are also much more fluid, this type of clothing thus offers less resistance than a shirt. How to choose your woman blouse? In order to choose your blouse for women, many elements must be taken into account. Here is a complete guide below. Collar and wrists Buy Now For the collar and cuffs of your blouse, it's important to think about the stiffness you want. Indeed, the more rigid the collar and the cuffs, the more your blouse will have a very formal appearance. For a more casual look, choose a c...
Read more

Frames for Men: How to Be More Attractive With Glasses

Image
Do you need glasses to see closely? Surely you have enjoyed a perfect view throughout your life and now after 50, you need glasses . If you think that they do not feel you or that they will make you look older, do not miss this article where we will give you some simple tricks to choose the glasses that best suits the shape of your face. Just like a good black suit makes you look slimmer and stylish, the right frame can minimize or emphasize the features of your face, such as a nose, forehead or pronounced chin. Next, we will deepen in the following subjects: What are the best glasses for men? Combine the color of your glasses. Aviator Glasses What Are the Best Glasses for Men? The glasses should contrast with the shape of your face, this is the golden rule. This means that angled frames will narrow a round face, while around glasses can soften a pronounced chin. If your face has a heart shape or inverted triangle, opt for wider glasses at the bottom, this will compensate ...
Read more

Hot Style Men New Hoodie, Sweatshirt Fashion 2019 Winter

Image
All that Paul Newman wore for six decades is programmed admission to the Hall of Fame Menswear. You don't have to see his papers, waves directly through. Newman adored a sweatshirt. He wore a BCBG like a symbol with blue eyes in the 50's and 60's and never took any appearing of honorary pathway occasions in his 80s as yet wearing the immortal style. Newman comprehended that in a sweatshirt, it may look easygoing yet thought to be chic, yet energetic. All together, all with a straight draw. It is seasonless, complimenting to all body shapes, simple to layer: one uncommon certified Everyman components of menswear. Buy Now Its intrigue is down to solace and flexibility, says Pelle Lundquist, organizer and inventive chief of the Swedish brand one day in March. "It's elusive progressively open to attire that goes well with what a great many people like to wear. A sweatshirt is anything but difficult to utilize both for a road or a chic look and easygoing "....
Read more