Since the start of the digital revolution, more than fifty years ago, the quality of machine translation has largely improved, and, nowadays, it is considered by many people a good way to overcome linguistic barriers and to transfer culture easily.
Indeed, at times, it happens to hear some people complaining about the prices of human translation and, consequently, deciding to rely on a computer software, like google translate or bing translator, to translate their texts into another language for free.
Luckily, in some cases, they manage to discover before publishing, printing or delivering their texts that the quality of such a translation is not as good as they expected.
This fact often surprises lay people, but this is no news for professional translators. In fact, as much as machine translation has improved, it still presents some important limitations.
Let’s look together at the pros and cons of machine translation:
- Faster than human translation: you can do a translation with a click.
- Cheaper than human translation: it has no costs.
- Everyone can translate a text: even people who don’t know Italian can translate a text into Italian.
- Multi-language translations: you don’t need to search for different translators, each one specializing in a specific language, to have a text translated into, let’s say, Italian, Russian, Spanish and Japanese. It translates into any language.
- Accuracy: the output generally presents semantic and syntactic problems.
- Accuracy: the outputs quality changes according to source and target languages – i.e. Italian > English translations are slightly better than English > Italian translations.
The positive aspects of machine translation are remarkable. I mean, the possibility of translating into every language in a flash, and with no costs, ideally makes machine translation one of the most powerful instruments for global communication.
However, these positive aspects are worth nothing if the translated texts lack in accuracy. Indeed, without a correct translation of both words, terminology, grammatical structures, and idiomatic expressions, a machine translated text may really undermine people’s understanding of it. For an author who decided to use machine translation to translate his/her book, this would mean receiving horrible reviews.
Maybe at this point someone could think I’m exaggerating. So, to clear any doubt, I run a series of tests – both with literary and technical texts. Here I’m going to show only two of them.
I put an excerpt taken from a users’ manual and an excerpt taken from Wuthering Heights into google translate to get their Italian translation. Below are the results:
The translation made by google translate of text 1 presents some understandable sentences – the last two. However, their translation is grammatically incorrect and stylistically unsuitable for a product’s manual – i.e. lack of use of the imperative tense.
In addition, the translation of text 1 also shows some significant terminological problems – i.e. harness clamp becomes morsetto cablaggio instead of morsetto dell’imbracatura.
If such a translation was printed, included with product and, then, delivered to customers, it could have caused understanding problems, or even lead to people getting hurt. For the firm, this would have meant legal problems, and money and customers’ loss.
The translation by google translate of text 2, instead, has just one exact sentence translation1 – Questo è certamente un paese bellissimo! All the other sentences are simply incomprehensible.
If Emily Brontë had lived in these times and had published such a translation, I really doubt that Wuthering Heights would have been considered a masterpiece in Italy.
These two examples show that even if machine translation has improved a lot since 1900, its accuracy still is light years behind that provided by human translation. This is because human translators have at least one quality that machines lack: interpreting. We do make translation choices based on grammar and vocabulary, but we do so while considering, among others, style, context, logic, and target culture.
Google translate, and machine translation in general, is a good instrument for personal – not-professional – everyday situations – i.e. to communicate with friends, or to try to grab the global sense of a foreign text.
However, if you need a good translation for your profession – business, books, manuals, etc. – you’d still better search for a professional translator or, at least, a native professional that check the accuracy and suitability of the machine translated text for you.
Maybe, in the future machines will be able to offer a “good” word-for-word translation but, in my opinion, they won’t ever have that knowledge of the world, typical of human beings, that is essential to carry out understandable and high quality translations.