On the one side, there are the technical translators or translation agencies. These guys sink their teeth into manuals and operating instructions for everything from medical equipment to food processors. Purely logical content and plenty of jargon. These kinds of texts can only be interpreted in one way, with no scope for nuance. To put it simply, this is all about translating words. More recently, hybrid engines have started to produce more acceptable output, but don’t get your hopes too high. If the same output came from a human translator, it would be frowned upon and classified as poor.
A long way to go
It goes without saying that this technical segment is more prone to disruption, even though technology has been integrated into the translation process for quite some time. Machine translation is not a swear word in this part of the translation world. That being said, a human revisor is an absolute must. Unless you’re aiming for some first-class comedy, that is! A recently published poll on ProZ.com – the world’s largest online community of translators – revealed that only 4% of all translators have no reservations about checking machine translations. The reason? The quality is just too low. Disruption is on its way – incorrectly and hilariously translated by Google Translate as ontwrichting is op zijn manier in Dutch (= disruption is on its manner) – but it still has a long way to go.
The other side of the translation world is where the non-technical translators live. Be it commercial websites, pure storytelling or marketing communication, the big difference here lies in the fact that it is ideas which are translated rather than words. You could describe technical translation as a skill and creative translation as an art. I am therefore convinced that computers will never be able to compete with creative translators. Not now, and not tomorrow either. Self-learning or not.
Translation = writing
You might be able to teach a computer vocabulary, grammar and spelling, but what about tone of voice? What about style? And one step further even, what about language humour or irony? It is precisely this emotional charge which renders an algorithmic analysis or approach to language impossible. The translation of technical content will undoubtedly continue to improve, but what about ‘real’ language? How would MT handle an ambiguous phrase such as “Time flies like an arrow; fruit flies like a banana”, for example? As argued above, you can teach a computer how to write correctly, but it will never win a Booker Prize. The same applies to translation, as translation is writing first and foremost. “Writers make national literature, while translators make universal literature”, as José Saramago put it so neatly.
If MT ever manages to pass our translation test – more treacherous than Mordor, according to insiders – I would be prepared to review my position. And quite possibly to eat my hat. In the meantime, let me sum up with an infamous translation quote: translators who can be replaced by machines, should be.