Hannah is a 31-year-old translator from Birmingham. This interview was recorded over Zoom. It has been lightly edited for structure and concision.
I have been a translator for seven years. Technology has had a massive impact on the amount of time I spend working. In fact, technology has had a massive impact on translation across the scale.
When I was at university doing my Masters, our professors told us about this amazing new world of translation tools. They were presented to us just as tools to help us along. And initially that’s what they felt like. But in the past two or three years, agencies have been demanding that you use specific tools. You won’t be given the project if you don’t use the specific computer automatic translation tool they ask you to use.
Now, clients no longer want you to send back just the translation, but what’s called the translation memory. A translation memory is like a massive database. What that means is that when you use the software, you open up the document to be translated, and alongside that you open up the translation memory. So as you type, as you save your work, and ultimately when you deliver, everything you’ve translated is not only in the file, it’s also now included in this translation memory.
The crazy paradox is that by giving back the translation memories, you’re essentially putting yourself out of a job. Because you’re forced to use the software, you’ve not only given them the translation, you’ve also contributed to their automated database.
There are clearly two schools of thought about this. The old school are the translators who think this is absolutely appalling, that you can never replace human translation, and they refuse to work with it. Then you’ve got the new school, usually the younger crowd. They’re tech-savvy and they love this.
The big thing that everyone is talking about is called DeepL. It’s basically Google Translate taken to a new level. This is revolutionary. There’s no software to download, you just type in DeepL.com, copy and paste your text and out it’s spat into whatever language you want. Apparently this is eerily accurate. Even for more complex texts that would traditionally be felt to need a human touch. So now what’s happening – and this is really getting the backs of the old guard up – is that people are going onto these platforms and job boards like Fiverr. They’re taking on translation jobs for what I’d consider crazy money, 0.02 cents per word. They’re snatching up these jobs for cheap, jaunting over to DeepL.com, pasting the text in there, proofreading it quickly, and sending it off.
There’s a new thing happening called PEMT. PEMT stands for Post Editing Machine Translation. Sounds really dystopian, right? This is where a piece of software uses the translation memories—the database we’ve all contributed to—and automatically generates a text. And my heart sank when one of my favourite agency clients told me, “we’re going to be doing a lot more PEMT, can you let us know you’re OK with that?”
When I opened up my first PEMT project, the rate was so much lower than a translation. Initially I thought “I can’t believe this, I bet the text is going to be full of mistakes, I’m going to spend loads of time cleaning up — more time rejigging a badly translated computer sentence, than if I were just translating”. Then I opened it up, and once again I was simultaneously excited and dismayed to see the software had done a really good job.
In the future I think there are going to be two categories. One which is post-editing – cleaning up after the machine – one which is hyper-specialisation. And hyper-specialisation means being a legal expert who also happens to be a translator, or being a medical expert who also happens to translate. Because these are ‘high risk’ translations, they can’t really afford to automate them.
When I was doing my Masters, we learned translation is a craft and an art, finding just the right word to capture the exact atmosphere and meaning. That’s why it’s sad for lots of translators when we see that shift toward technology, because we lose a part of that art.
There are a few cynical people out there who say, ‘oh but people have always been worried about being replaced. That happened during the Industrial Revolution. But we’ll always need people to supervise the robots.’ The problem is that doesn’t apply to our situation now. Because when people supervise the machines, for example in my case, PEMT – what will happen when technology is created to do the PEMT? The machines are becoming smarter and smarter, which means the people supervising the machines will need to be smarter and smarter.
We’re going to reach a situation in which tech is so omnipresent and intrusive that the only people capable of cleaning up after the machines will be web people, developers, coders, engineers. So what are the masses going to do?
This is going to be a really interesting conversation across all sectors and all professions. How do you define yourself? So many more jobs are becoming automated that I don’t think you can any longer define yourself by your work. As jobs start dropping off, and tech takes over, and now, especially with coronavirus.
The debate around a citizen’s wage and universal credit, this new utopian way of how we could live – maybe I’m overly optimistic, but I think this pandemic we’re having right now might rejig conversations about it. I think people are going to be looking for ways to get fulfilment outside their work, forcibly so. They tell us that we’re now staring down the barrel of a financial crash similar to the Great Depression. A lot of people will get laid off. What’s going to happen in the future? I think companies are going to turn towards things like DeepL, because a crash is coming. And a crash means companies are going to do everything they can to do stuff on the cheap. Maybe faced with this, governments will start talking about the possibility of some kind of citizen’s wage.