By Caroline Lemineur, translator at Right Ink

 As globalisation shows no sign of slowing down, there aren’t many people who have never had a go on Google Translate. Actually, machine translation is nothing new: did you know that it was first trialled in the 1950s during the Cold War? Back then, it was used by the two big superpowers, the Americans and the Soviets, to translate intercepted telephone calls. The initial results were promising, but turned out to be less than conclusive. So why are we talking about it again now? Because in recent years, the translation industry has been going through a bit of a revolution thanks to the incredible progress made by the arrival of neural machine translation.

NMT to those in the know, neural machine translation is a system that has been developed relatively recently, based on networks of artificial neurones and very (very!) large volumes of data. The famous big data. This system uses a huge amount of computational power, but it’s a long way away from being infallible. Indeed, while it copes pretty well with the vocabulary and grammar of its working languages, it sometimes trips up on syntax and doesn’t really understand what it’s translating.

And that’s the problem! The meaning of a phrase — and all the more so of a text — doesn’t just depend on its individual words or the bits and pieces that connect them; it also depends on the context, its author and its audience. So the word “bark” means something different depending on whether you’re a vet or a botanist. On top of this, languages are full of wonderfully rich phrases that are totally unique. For example, the Swedish expression “jag sitter i tankar” literally means “I’m sitting in my thoughts”. A translator with any common sense would no doubt render this as “I’m lost in thought”. DeepL, one of the best translation tools currently available, insists on suggesting “I sit in tanks”. Hmm, this is getting complicated…

Thankfully, machine translation isn’t a lost cause. Far from it in fact. The examples given above should nevertheless encourage you to be a bit wary, and make sure you call upon the professionals, who understand the potential pitfalls of NMT and know how to overcome them. Indeed, while computers may be able to translate words, the experts know how to translate ideas…

So please feel free to get in touch with us to find out more about the opportunities and the hazards of this translation tool 2.0. Our experts at Right Ink are keeping up with the latest developments in this field, and look forward to talking you through the potential benefits.