top of page
  • Writer's pictureSarah Hudson

Rage against the machine – a translator’s view of translation technology

There are moments when I feel like a 19th century cotton weaver faced by the arrival of the spinning jenny. Part of me wants things to remain as they are and I also want to make sure my skills will still be in demand, valued and fairly compensated. But like it or not, the machines are coming. Machine translation (MT) tools such as Google Translate, DeepL and PROMPT Translator are ushering in a new age of quick, free and accessible translation. This has got to be a good thing, right? Anything that enables citizens of the world to communicate more effectively wins my vote. It has been wonderful to hear how MT has helped aid workers communicate with refugees in war-torn regions and how it united football fans during the 2016 FIFA World Cup. Maybe the future that Douglas Adams envisaged in the Hitchhiker's Guide to the Galaxy with his Babel fish instant translator will soon come to pass.

It is very tempting for freelance translators to recoil from these developments. We have worked hard to acquire our skills, only to feel that the rug is being pulled from beneath our feet. Our own unconscious bias automatically makes us home in on the limitations of MT and gleefully point out embarrassing machine mistranslations such as Benjamin Netanyahu’s infamous tweet to Eurovision winner Netta Barzilai. Instead of translating the Hebrew word kapara as “Netta, you little darling”, the MT tool used by Twitter rendered it as “Neta, you’re a real cow”. Ooops! Developers of this technology are acutely aware of the potential for errors. In fact, the disclaimer on the Google Translate website warns users against blind acceptance of the machine output. Nonetheless, huge corporations such as Amazon and Baidu still believe in the potential of MT and are pouring massive amounts of money into its development.

I often get asked whether I’m worried if Google Translate will put me out of a job. My answer is that I don’t believe I will have to hang up my translator’s hat anytime soon and that I am focusing on the benefits that MT has to offer. For example, it’s great to have the machine do all the mundane tasks, such as automatically search databases and dictionaries for definitions, synonyms and specialist terminology and provide suggestions. I am therefore very happy to embrace something that has the potential to improve my productivity, assist with quality assurance and even support creativity. The latest refinement of MT called Adaptive MT also holds a great deal of promise as I believe it aims to put the human translator at the heart of the process by providing greater control over editing and allowing the user to draw upon a variety of resources.

So it’s very unlikely that I will be brandishing a pitchfork and calling for an end to machine translation as I can see that it can offer ways to make me more efficient and creative. On the other hand, I will still retain a healthy degree of wariness when it comes to some of the exaggerated claims being made by certain technology companies about the brilliance of MT. It is going to be a long time before machine translation will match human translation and here’s why:

  • Language is what sets humans apart from every other creature on Earth. Our highly-evolved and complex brains enable us to process words and meaning in a way computers are unlikely to replicate for a long time to come.

  • Computers are inanimate and devoid of emotion. It is hard to imagine them being able to use language in a truly creative and evocative way.

  • Language is dynamic and changing on a daily basis in accordance with socio-political developments. It will be difficult for machines to keep pace with this.

  • Whilst producing some impressive results, machine translation still has significant limitations in terms of accuracy and its ability to understand context and produce natural-sounding language. Added to this is the fact that it can only process common language pairs at the moment, so there will definitely be a need for Swahili to Welsh translators for a long time to come.

  • I believe that machine translation is going to be the equivalent of mass-produced products. There will still be a demand from translation buyers who appreciate the craftsmanship involved in high-quality human translation.

I would like to end with a thought-provoking quotation that was included in the Tool Box Journal, a very informative journal on translation technology published by German–American translator Jost Zetzsche.

"Words are not coins, dead things whose value can be mathematically computed. You can't quote an exact English equivalent for a French word, as you might quote an exact English equivalent for a French coin. Words are living things, full of shades of meaning, full of associations and, what's more, they're apt to change their significance from one generation to the next. The translator who understands his job feels, constantly, like Alice in Wonderland trying to play croquet with flamingos for mallets and hedgehogs for balls; words are forever eluding his grasp."

Ronald Knox, British crime writer and translator


bottom of page