Don’t get me wrong. As a technophile, I’m all for technology when it comes to increasing efficiency and improving lives. Machine Translation was already a ‘thing’ when I studied for my Masters back in 2008, but since then it has boomed, with many new startups on the scene, and established translation businesses such as SDL, Systran and Google investing heavily in Machine Learning.
Used correctly, in organisations that produce huge amounts of content on a regular basis, Machine Learning technology can do a lot of the leg-work when translating standardised, predictable content. Many large organisations have invested time and effort in building up a corpora of translated content, which in the long run will help them gain efficiencies and ensure their content uses consistent language that is faithful to their brand voice.
However, like with all technical solutions, without the right application and human oversight, Machine Translation can fail dramatically. The recent story of the Osaka Metro having to shut down the English version of its website due to a translation blunder just goes to show that cutting corners and relying purely on machine-generated translations can go terribly wrong, especially on such a public platform like a website.
For SMBs, and for highly public or litigious industries, human involvement in the translation process is imperative. A website is your medium to promote your organisation or provide important information to your audience, so why would you risk damaging your reputation by publishing nonsense? Similarly, in the legal or medical professions, for example, having 100% accurate documentation produced by a specialist in the field is crucial.
So what is it that humans can do and machines can’t? Quite simply, they can understand context and linguistic nuance much better, and apply this knowledge to the translated content. They take into account who is writing, who they’re writing for, what has been said before and whether the writer is using an idiom or subtle wit to get their point across. Humans can adopt a particular tone of voice and appropriate level of formality, and break down complex grammatical structures found in languages such as German, to produce a piece of writing that reads like it was originally written by a native speaker.
When it comes to translation and localisation, it’s important to understand how humans can work with technology to produce the best outcome for organisations and their audiences. More about how translators leverage translation software to maintain quality and provide a cost-effective service in a future post.