Machine translation: Babel or babble?

Translation agencies relish it, professional translators tremble at the thought. So, what do machine translations have to offer, and what not?

Let’s start at the very beginning. Before the 1990s, if we wanted to translate a text, we had to consult professional translators trained to use dictionaries and choose the right words. Indeed, since a word often has different meanings, we would have to use the context to find the right one.

This was a time-consuming job. So machine translation was introduced. Getting a quick and cheap translation was the motto, but the result was often very poor.

Which is why the typical word-for-word translation technique was abandoned. Machine translation became a combination of pure machine translation and human translations. Google Translate’s database, for example, is filled with translations that have been produced by humans. Take away the human contribution, and it becomes a totally useless tool.

However, it is not because human translations are used in machine translation that machine translation suddenly becomes a valid alternative to pure human translation.


Translation agencies that use machine translation often talk about post-editing: the work of translators who correct the machine translation and receive at best, half their normal translation rate. What these agencies do not mention, is that such translations are of poor quality even after the correction. The translators in question, who are often not professional translators and only hired for their knowledge of a language, are always coerced into a specific style by the machine translation: the machine’s style. Their task mainly consists of correcting language errors. If they were to translate the text from scratch, the translation would look a whole lot different.


Machine translation has everything to do with money and nothing with quality.

We also have to remember that machine translations increasingly receive less post-editing due to the added human effort required. Raw machine translation has become an easy means to earn money. Software for machine translation is also available for professional translators, often for free. They only need to invest the time needed to use the software. Yet, professional translators do not do so. With good reason: they prefer quality to quantity. Their quality is exactly what distinguishes them from those who do make use of machine translation.

Over time, machine translation with minimal human input via post-editing may indeed produce an understandable translation, but not necessarily a good translation.

This, of course, does not mean that there is no market for machine translation. Industries that always use the same types of texts (e.g. IT and automobile industries) – where texts of excellent quality are less important and understandabililty is enough – might make use of machine translation. Yet, there are limitations: the terminology used must not be ambiguous, there is no room for idiomatic expressions and the goal cannot be a perfect result. That means, for instance, that legal and marketing documents are, from the very start, destined to fail if they are translated using machine translation.

And it does not mean either that professional translators cannot make use of machine translation as a supplementary tool. But they need to know very well what they are doing and not fall into the various traps.

Problems related to the use of machine translation

It is often forgotten that machine translation assumes that the source text is of impeccable quality. Anyone with experience in the translation industry knows this does not happen often. Bad grammar, spelling errors, missing words, incomplete sentences … occur often. A translator can spot and correct these errors and still produce a great translation. Unless of course, unqualified translators are hired to try to correct a machine translation at a very low price, resulting in chaos.

To say nothing of the differentiations, subtleties, sarcasm, idiomatic expressions, ambiguities, puns, word colouring, tone… all of which you will find in texts on a daily basis, but which machine translation cannot handle, and never will be able to handle. Language is simply too complex.

There is also a cultural aspect. There are countries where badly translated texts are easily accepted. However, in countries such as Canada, Switzerland and of course Belgium, people are less tolerant. Badly written documents are a disaster to a company’s image. Machine translation is not much different.

Three factors

In the translation industry there are three factors to a translation: speed, price, and quality. Optimising two of these factors together is possible, but never all three factors. The same applies to machine translation: it’s quick, it’s cheap, but the quality is not good. Those trying to convince you of the contrary, are engaging in marketing gibberish.

Apart from the quality issue, there are other considerations that need to be taken into account when dealing with machine translation. Think about confidentiality. What happens to your documents that are translated by a translation agency with machine translation? You have absolutely no idea nor control. This doesn’t mean that your documents are being distributed, but the risk is much greater than with a professional translator who works ‘serverless’ on his or her own computer.

Companies using their own database but also external services such as Google Translate, not only receive information from these external services but also send information to them. Confidentiality is not always guaranteed!


When a company receives a translation that was produced using machine translation, does it have a copyright to the translation? Copyright is only applicable if the work is the result of a creative effort and is original. In the case of machine translation, there is no creative effort, let alone an original work. Therefore, there is no copyright.

A company that uses machine translation has no legal ground to prosecute people who use the machine translated texts without credit or permission.

I am presenting the issue in black and white here. But, when considering the use of machine translation, all of this information needs to be taken into account.


If you want a high-quality translation, you need a professional translator. Quality costs money and time.

Machine translation can, at best, lead to ‘understandable’ translations. It is the end customer’s responsibility to determine which factor is most important: speed, price, or professional quality.


Machines translate words, translators and interpreters translate language.

Would you like more information?