Let’s be honest: I bet every single one of us has used at least once a machine translation programme, such as Google Translate or DeepL, to translate a sentence or a text. These programmes come in handy when you want to get an idea of the content of a text, but that’s all there is. Their software is not perfect yet and cannot produce flawless texts at all: often there is no consistency, sentences are being misinterpreted or texts contain strange sentences or even translation mistakes. And these can have disastrous consequences.
First, a translation machine does not necessarily adhere to the terminology the client uses. Given that some words have different translations, it is of the utmost importance that in documents for one and the same client the same translation is used at all times. Translation machines are aware of these different meanings, but they are not always able to choose the right one.
Example: the English word ‘employee’ can be translated into Dutch as ‘werknemer’ or ‘medewerker’. Every company has its own style guide, which should be followed at all times.
Ideally, translation machines should also be able to determine the right register. In English, the pronoun ‘you’ can be used both in formal as in informal contexts. In Belgium or in Germany, it is very important to choose the right register (‘u’ or ‘je’), because the wrong choice can be regarded as disrespectful or inappropriate.
Example: if you address someone in Dutch with ‘mevrouw’ (‘Miss / Madam’) or ‘meneer’ (‘Sir’), you should use the pronoun ‘u’, not ‘je’.
Moreover, translation machines are not capable of detecting errors in the source text. As a professional translator you read the text, you understand it and think logically. When a word is missing or the text contains strange connections, you will be aware of them. Translation machines, however, translate those texts blindly, without really thinking about them.
Example: in a fire safety brochure, the following is stated: ‘Make sure to stick your hands in the fire’. In this sentence, it is obvious that the word ‘not’ is missing. Translation machines, however, do not notice that.
And let’s not forget about the language and translation mistakes of translation machines. Languages evolve constantly, but machines cannot always keep pace. New words are often not recognised or words that have evolved over time are still being translated the old way.
Culture might be even more important than the insufficiencies on a linguistic level. Every culture is unique and has its own expressions, slang or names. These can be a real obstacle for translation machines. Professional translators, on the other hand, have a good grasp of the target language and are perfectly equipped to translate those cultural concepts.
As a matter of fact, some languages are spoken in different countries. And they are not even that hard to find. Just think about the difference between Belgium and the Netherlands. Some words and expressions are much more common in Belgium than in our neighbouring country. Translation machines are just not capable of making that distinction.
Example: let’s take the simple word ‘Hi’ in English. Google Translate and DeepL translate it as ‘Hoi’, a translation that sounds very Dutch (Netherlands) and is not that common in Belgium.
Example: in the Netherlands they say ‘mondkapjes’, while in Belgium they use ‘mondmaskers’ for the English word ‘face mask’. Still, both DeepL and Google Translate translate the word ‘face mask’ as ‘gezichtsmasker’. However, this translation is incorrect, since it has a different meaning. ‘Gezichtsmasker’ refers to a cosmetic mask you put on to relax or to take care of your skin.
Let’s put these translation machines to the test and translate some sentences with Google Translate and DeepL.
The above excerpt is taken from the General Data Protection Regulation (Regulation (EU) 2016/679). Many privacy statements and data processing agreements are based on this regulation. However, Google Translate’s translation does not correspond to the official translation on the website of the European Union.
When translating marketing texts, translation machines fail unmistakably. Both translations of Google Translate and DeepL do show what the author of the texts wants to convey, but the translations are not attractive at all and Dutch native speakers immediately know something is wrong and notice that the text is not written or translated by a native speaker. This could severely harm the reputation of a company.
The English expression ‘to go into labour’ is translated incorrectly. DeepL translated this sentence literally, thus significantly changing the meaning of the sentence. In the Dutch version, it is stated that she started working on the plane. This shows that translation machines have no ‘feeling’ for languages.
Also slang appears to be an obstacle for translation machines. English speakers will immediately know this sentence does not make any sense.
It is clear that the quality of machine translations can have serious consequences for a company’s reputation. If you blindly copy the translations that automatic translation programmes such as Google Translate or DeepL generate, you risk losing customers, since they will lose all faith in your company. After all, with these low-quality translations you basically show them that you did not even make the effort to translate your texts decently.
In conclusion, it is of the utmost importance to entrust your translations to a professional translator. Only use machine translation if you want to get a quick idea of the content, and even in those cases, always be aware of potential mistakes!
Do you have a translation project for us?
Feel free to contact us.