TRANSLATION | TECHNOLOGY

I Never Wanted To Be a Translator Anyway, So There!

My uneasy relationship with language technology

No, I never wanted to be a translator. Nor did any of my cohort at school or university. Anyone choosing to study languages quickly became accustomed in my day to a fairly standard response:

“Oh.” (A slightly embarrassed or puzzled look flickers across their face for a moment.)

“So I suppose you plan on being a teacher? Or maybe a translator?”

At which point you were already wondering whether it was even worth explaining that your passion in fact lay in mastering the intricacies of Middle High German and embarking on postgraduate research into Hartmann von Aue’s 12th-century epic romance Erec (coming soon to Netflix: a hack-and-slash fantasy starring Henry Cavill as the oversexed knight errant — or maybe not).

Translating for a living seemed like an admission of defeat, an acknowledgement that the only purpose of studying another language was to serve as a paid go-between, dutifully trotting back and forth from one linguistic camp to another, carrying someone else’s messages in a satchel.

But become a translator I did. The gateway drug, as for many of my colleagues, came in the form of the odd side gig while working as a language teacher (yes, I ticked that box as well — they turned out to be right all along…). Before long, translation took over, the work flowed in, a mortgage and kids followed, and the rest is a history of decently paid but ultimately frustrating routine.

Don’t get me wrong — translation has its fascinations. Working across a range of sectors, at its best any given day can feel like an eye-opening series of episodes of the Discovery Channel’s How Do They Do It?, a behind-the-scenes insight into the intricacies of other people’s jobs and processes I had never given a second thought to, or even knew existed.

The integrated water cycle, the intricate to-and-fro of payment data when you refuel your car at an automated pump on a highway in Costa Rica, the different specifications for crew and officer toilets on board a refurbished tuna-fishing boat (stainless steel for the lowly crewmembers, porcelain for their superiors, in case you were wondering).

The translator’s life hasn’t exactly been a wild ride, but it has certainly given me a glimpse into more aspects of our world than I might have imagined when I first tapped away at my keyboard all those years ago, to produce an English-language version of a Spanish non-residential lease agreement.

Now, though, the road seems to be running out. It is only a few years since Yahoo! Babel Fish (remember that?), Google Translate and other such machine translation services were more often than not an excruciating embarrassment, the object of snarky jokes and abashed apologies. The finest brains of the computational and linguistic sectors had been slaving away since the 1950s, and come up with nothing better than a flightless pidgin.

Not anymore. Everyone’s favourite bugbear/genie-in-a-bottle/Next Big Thing, AI, or in this case more specifically neural machine translation, NMT, has been fed billions upon billions of (other people’s) words well after midnight, has been wetted with billions more dollars of investment, and become a different beast altogether.

Ubiquitous, voracious, but above all, actually very good. Good enough to pinch maybe half my workload from right under my nose. Above all boilerplate, bread-and-butter stuff at which it excels. Sure, unless the client wants to entrust its reputation entirely to the inscrutable judgment of the daemon that lives inside the black box, a flesh-and-blood (and eyes-and-brain) translator is still needed to post-edit the output. But the whole point of technology is to reduce the human workload. Which inevitably means either less work, or lower rates of pay, or both.

And it is remarkable how much and how rapidly the recent ChatGPT media frenzy seems to have convinced companies, and their bean-counters, that they can slash a major cost item from the budget of producing and publishing that annual CSR report for their multilingual stakeholders (who maybe only ever skim the executive summary anyway). The old Babel Fish joke is now on us, the human translators.

After a few months of growing panic, then, I have decided to embrace the change, to make use of the downtime in my working day, and return to what I always wanted to do, back in my wide-eyed youth, before the mortgage-slave drudgery struck: to write my own words rather than render someone else’s.

I find the whole business of Large Language Models intriguing and beautiful, as well as threatening (Tyger, Tyger, burning bright / In the forests of the night / What immortal hand or eye / Could frame thy fearful symmetry?), and hope to explore that brave new realm, while also pursuing other cultural interests and curios. Spain, where I live, is a remarkable and still surprisingly misunderstood country, so I will be trying to get out from behind my desk and explore and explain its more obscure delights, as well.

So thanks, AI. You stole my lunch but gave me back my lunch hour. And my hunger. Though not literally, I hope.