Is it just me, or does it seem like every conversation in the localization world these days comes with a side of machine translation and AI? It's like I'm back in middle school and MT is Backstreet Boys' "I want it that way" – and if you think this analogy is stupid, you should have seen what chatGPT proposed I'd use.

But somehow, through all that chatter, MT for UX still remains mostly undisturbed, in the sense that people are completely ignoring it. Try and find an MT report highlighting UX or UI as one of its domains - I dare you, because you won't, because there isn't. Everyone's elegantly ignoring UX, since using MT or AI to generate passable-quality multilingual UX copy is HARD, y'all. Some would say, so hard that it's not worth the effort.

I'd like to question this premise – a hobby of mine, along with reading fluffy novels and hunting for scones (we're not big on scones in this country). I still think we can get passable results using MT for UX, and more than that, I think companies that use this time to experiment and learn will be the first to benefit from MT when we do get translation singularity*.

And yes, plenty of companies are already getting some mileage out of MT. Mostly, they run their content through MT first, then pass it over to their human experts for post-editing (PE). Even if they focus mostly on none-UX content (help centers, blog posts and more) it still means that both loc managers and linguists get to practice MT-driven processes.

If you'd like to learn more about this super interesting topic, join Boryana Nenova and yours truly 🙋🏻‍♀️ on June 20th, at 10:00 CET for a deep-dive webinar on machine translation for UX. There will also be a recording for those who can't make it live - but still, make sure you sign up to get the link when it's up!

*Translation singularity was all the rage a few months back when Translated announced we're steamrolling into an age where we won't really need humans for translations anymore. I like to imagine we'd all be sitting in our gardens then, eating scones and just overall enjoying life. My kids will be older by then and will obviously be serving me fresh coffee whenever I snap my fingers. Reality's great, isn't it?

How is this still happening?

Some companies, in an earnest (although misguided) attempt to save costs (which I get), play a peculiar game of hide-and-seek with their legacy content. How? By excluding it completely from the word count, meaning translators are expected to localize new content without any reference to the existing history. It simply doesn't make sense. I’ve used my significant illustration skills to make this very clear:

Basically, every part of the screen is translated in a different way. Maybe one is singular, one is plural. Maybe sometimes the product name is left in English and sometimes it’s translated. Who knows? It’s a wild west of linguistic chaos. The result is that we end up with this awkward Frankenstein-ish blend of localized content. And other than the fact that it’s disorienting for users, it’s also completely disrespectful. Do you want your users in other markets to feel like they’re an afterthought? Because that’s a surefire way to make them U-turn straight to the competition at the first chance they get. This is a rant, but it’s also me begging you to be proactive about this. If you’re a linguist and you get a request to overlook 100% matches, make sure your client knows exactly what they’re asking - and exactly where it might lead. And if you are a loc PM… Please, just don’t, OK? I’ll send you a scone.