Why Machine Translation is not enough

Robot and girl representing machine translation and reviewer

We have already discussed in one of our previous articles how localising your website can boost your exports and increase customer satisfaction. In order to achieve these results, you may simply translate your website using a machine translation plugin. However, is Machine Translation (MT) adequate for your website localisation?

Machine Translation History

Taking a step back, we should check Machine Translation’s history and discover the steps that led to the development of such a mechanism. The concept of easy transition between languages appeared in the 17th century when René Descartes expressed his thought of a global language. Three centuries later, Machine Translation research programmes appeared, relevant conferences were held, until an MT system translated military manuals during the Vietnam War. Since then, Machine Translation has been broadly used both in the academic and commercial fields.

Why use Machine Translation

The main advantage of MT is that it offers fast, easy and inexpensive translation. Those benefits are quite appealing for a business owner who wants to save a few pennies on a seemingly simple website translation. In fact, there are several businesses that have chosen to go down that road, despite its shortcomings.

Mistranslations

If we put aside those obvious benefits, a few points need to be considered. First of all, conventional machine translation cannot distinguish the different meanings of the same word, which are dictated by context. In addition, proper names can’t be easily identified and properly translated by Machine Translation. Syntactic issues may also arise, especially between languages whose syntactic rules are entirely different. The same applies for grammatical issues. In some languages the articles, adjectives and verbs are adjusted according to the genre of the object or person and verbs change depending on pronouns.

The Importance of the Human Reviewer

These are only a few aspects of translation that can go wrong when Machine Translation is used on its own. MT is also not able to choose the correct register or follow your company Style Guide. It can only translate word for word and cannot produce naturally flowing marketing text, appropriate for a customer facing website, or a digital campaign. When it comes to SEO, Google and its partners punish pages that include poor machine-translated content, meaning that your website can be led to a significantly lower position.

To wrap it up, machine translation may speed up the localisation process and reduce your spending, but it’s not appropriate for every type of content. Trained neural MT can produce satisfying results for technical guides for certain language combinations, but in order to achieve compelling and engaging customer facing texts, MT results should always be reviewed by a specialised translator. Although there has been significant progress in machine translation, a person’s contribution has not been replaced and is essential for a proper translation.

In case you are interested in using Machine Translation for localising your content, Nexya can help you navigate this complex process and choose the right solution. Contact us to discuss your needs in detail.