Machine Translation: Friend or Foe?

Now is an exciting time for the translation industry. With such massive technological developments taking place year after year, the traditional image of a translator sitting at their desk, sheets of paper everywhere and a well-thumbed dictionary to hand is, without doubt, a thing of the past. Nowadays, it’s all cloud-based glossaries, real-time translation memory updates and collaborative, international, interconnected working…

But there’s one thing with the potential to change the face of the translation business more than anything before, namely – machine translation. What was once a highly specialised tool reserved for the most carefully selected projects and niche genres is now becoming part of the furniture across the board. Many people will already be familiar with the likes of Bing Translator and Skype Translator, tools that allow online content to be instantly translated at the click of a mouse when using platforms like Facebook or Skype, for example, and have meant the world wide web’s estimated 4.86 billion pages are now linguistically accessible to all.

At The Translation People, we understand the importance of embracing new technologies and the benefits they can bring – for us, as well as our partners. But we also place the utmost importance on the quality of our clients’ translations. So how can the two ever gel together? Can a machine translation ever compare to that of a living, breathing human translator?

Essentially, yes – but only with great care and consideration.

In certain cases, there is a real argument to apply machine translations to appropriate texts as part of a planned process. When customers need translations of technical documents, scientific brochures, legal material, and so forth, machine translation engines do exist that can impressively translate the source material into comprehensible translations almost instantaneously – but only after significant effort in training the engines and as part of a carefully planned workflow, neither of which are free. But if you add in a corpus of translations already saved in translation memories, combined with the vast amounts of material the engine has been trained with, getting a good translation from this carefully controlled reference material is a very strong possibility indeed and can certainly be worth the effort.

But that leads us nicely onto our next question – can machine translation be trusted to work alone?

As the technology stands at the moment, no. No machine is yet capable of producing a fully-fledged translation to rival that carried out by a linguist. No computer, program or plug-in, however many algorithms and reference material they’re fed with is capable of perfectly reflecting the nuances and subtleties of something as complex and changeable as the written word – yet.

So that’s where good post-editing comes in. Machine translation is capable of doing much of the groundwork, but only a great post-editor is able to turn that output into a great translation. In short, post-editors are fully-trained linguists who can efficiently sort the good from the bad, carefully comparing the automated translation to the original source text, checking and adapting what’s there to ensure everything is perfect and ready for the customer. They’ll work through the output, correcting issues such as word order, tenses, idioms, and so on. The framework will be there – which can save a great amount of time and effort on lots of projects – but only with skilled post-editing will the customer have a translation which is both ready for market and an apt, accurate representation of the source.

The greatest benefits of machine translation are becoming ever clearer to customers, linguists and language service providers alike as the technology grows and grows: greater speed, high quality and lower costs. So in answer to our first question – machine translation, friend or foe? – we have to say it’s undoubtedly our friend, so long as capable, trained and specialised linguists are there to lend a helping hand.

Read more