The Reasons why Machine Translation cannot fully replace Human Translators

The Reasons why Machine Translation cannot fully replace Human Translators

The Reasons why Machine Translation cannot fully replace Human Translators

Aucun commentaire sur The Reasons why Machine Translation cannot fully replace Human Translators

what-is-machine-translation

The Need for Technology in the Translation Sector

It’s a phenomenon you’ve probably noticed… Machine Translation has considerably reduced the language barrier. Most of your devices can now turn any word into any foreign language, both when entered as text or via voice recognition, thanks to Google Translate, Bing Translator or other tools.

For personal use, those technologies can allow users to roughly understand any foreign language and do their online shopping on foreign websites, or even talk to other people in different languages using real-time voice translation. But for business use, rough is not enough!

The emergence of Big Data, the mushrooming increase in content volume and Web 4.0 are the main drivers of the increase in translation demand, making translation technologies a must for translation companies to successfully meet the demand.

 

The Business Challenge of Content Translation

Budgets, staffing, content volume, time and a variety of other factors will consistently prompt organisations to shy away from translating even a small fraction of the content they have on hand. As a result, most information will never be translated into even one language, much less into many languages. That’s why companies like Amazon.com are using Machine Translation, given that the sheer volume of constantly changing content would overwhelm all the human translators in the world. However, they can use and customise their own Machine Translation engine since the content to be translated is highly repetitive. This is one of the numerous reasons why Machine Translation plays an increasingly important role in today’s translation industry!

 

How Machine Translation (MT) Works (in a Nutshell)

The two most common approaches to MT are Statistical Machine Translation (SMT) and Neural Machine Translation (NMT).

  • SMT is a software-based approach driven by Big Data (acquired from various sources) and statistical probabilities. In other words, this technology learns from grammatical and orthographic rules, and outputs translations based on the most common uses of words and expressions. This type of translations can seem very literal and can sometimes sound like they were done “word-for-word”.
  • NMT seeks to imitate the way that the human brain operates. By using deep learning technology, longer phrases can be more effectively translated. In other words, NMT will take into consideration the context of the statement and the idiomatic expressions used, in order to produce more advanced translations, by truly adapting a sentence’s meaning to another language (e.g. in English, we say “It’s raining cats and dogs”, while the equivalent French expression is “Il pleut des cordes”, which translates literally as “It’s raining ropes”). This example illustrates how the purpose of NMT technology is to not only translate the expression literally, but to make it understandable in other languages.

 

The Pitfalls of Automatic Translation

We can’t say it often enough—accuracy and quality are key requirements in a translation. A high-quality translation takes into account three key elements: the context of a paragraph, its meaning and style.

In a business context, misunderstandings due to poor translations can have serious consequences for your brand’s reputation with a knock-on effect on turnover.

According to a 2016 Shotfarm survey, 40% of online shoppers abandon their baskets because of poor product descriptions1. In addition, 1/4 of them returned products received because they failed to match the information in the product description. The quality of the product description and accuracy of translations is therefore vital. The sentences that make up a text aren’t independent from each other; the prevailing style and type of information are dispersed throughout. Therefore, unless this kind of information is memorised, it’s impossible to produce an accurate automatic translation.

 

Technology in the Service of Human Translators

A native-speaking professional can translate around 2,000 words a day; automatic translation is undoubtedly the hands-down winner in terms of response time. But as you will undoubtedly have realised, it can be very risky to entrust a business translation solely to a computer: the costs of a poor translation, although sometimes concealed, are nevertheless very real. More returned products which were bought online, damage to the company’s reputation, lack of professionalism, customer dissatisfaction… The list goes on and on!

Human translation is certainly more expensive than automatic translation, but the quality is infinitely superior. In addition, the impact of a good translation is considerable: a higher conversion rate, increased customer loyalty and better natural SEO on the various marketplaces and search engines…

Notwithstanding our previous comments, we don’t believe that there’s a war between machine and human translation. They are simply complementary approaches: automatic translation drastically reduces the time required to produce draft translations and thus optimises costs. It can provide a base for professional translators who can then rework the text, reformulate it, improve its writing style, and above all, localise it to suit the context and consumers in the target countries.

 

Contact us for more information about our Machine Translation services!

Leave a comment

You must be logged in to post a comment.

Réseaux sociaux – Suivez-nous sur…

Top 25 Language Service Provider in Asia in 2017

Contactez-nous

Back to Top