We’ve all seen lately what a huge leap forward machine-generated translation has made. Who among us did not ridicule Google Translate’s quality 10 years ago? Yet, progress has been steadily made by developers, and in 2018 it delivers widely acceptable translations by means of what is called “next-generation Neural translation technology.”1 It may still not be good enough for certain purposes, but we know too that other robust MT engines out there can actually handle those specialized, serious business cases we have in mind.
It seems people will be less and less needed in the world of translation. Translators especially can be worried about their fate. The blacksmiths of the 21st century – as Nataly Kelly put it in her Talk at Google2 – have crafted tools for themselves and others, but after a couple of decades, they will disappear from the scene. In the words of the speaker: “[…] just as blacksmiths helped move us into the industrial age, our modern-day wordsmiths – translators and interpreters – will help us move from the current age, in which information is merely available, to an age in which information also becomes highly relevant and useful.”3
At XTRF, we believe the PMs role should change as well, to overseeing projects that basically run all by themselves. Here is an ideal scenario we envisioned: (1) the requester drops files to a designated shared folder or sends them as an attachment to an email address, or orders a project via a portal; (2) files are then sent on to an engineer for preparation; (3) then converted to XLIFFs for all requested languages by a synchronized CAT tool; (4) translated; (5) reviewed; (6) quality checked and (7) delivered back to the one who initiated the whole process. We have pushed automation culture for years. In the words of a colleague of mine who works for an LSP, “PMs are like shepherds and projects are their sheep.” They should only be interested in the ones gone astray. A PM’s responsibilities in such a process are those of an Account Manager’s. The goal was to free up their time for more valuable work like building relationships with clients, or well, taking on more projects if needs be.
No language, no cry
It is estimated that 1 language dies every 14 days, so… out of today’s 7,000, half will not have any users by the end of this century4. It is the indigenous tongues that keep dying year-by-year. The last speakers of any almost-extinct dialect usually know one or two more commonly found languages that let them function outside of their family environments.
Such dialects remain an interesting piece for a scientific study or for hobbyists to pursue, but would never become useful for business purposes. Only 1% of all created text is translated today between dozens of the world’s most popular languages. If 90% of the world’s population communicates in one of just 25 languages5, why bother with a dialect spoken only by thousands? That must have been how Google decided to support 103 languages in their Translate service, by the way.
So it sounds heartless, but once a language dies there is one less language to compute. Perhaps 7,000 of them are simply too many for today’s world to communicate in? Ockham’s razor applies and causes my linguist heart to bleed… We need to come to terms with this uneasy truth that at some point, there will be fewer languages than there are today. It seems inevitable that computers and computer-generated texts will take over in the future. Money and time rule business, not sentiment. If machine translation will prove to be good enough and cost-effective, nobody will cry that a dialect ceased to be or that a translator in a distant land lost their job.
However, on the flip side, the globalization phenomenon is enabling certain languages to gain strength in numbers. And while English continues to dominate, it is by no means the only global language that matters, with Spanish and Mandarin Chinese now hot on its heels.
For translators, this may mean that demand for certain language combinations is likely to increase in the coming years, especially with business opening up in countries across the Asian continent6.
The good news is that in a world where machines understand people no matter what language they speak, our local dialects will be strengthened. Even artificially created languages such as Esperanto may benefit from such a fact. Who knows maybe even Sindarin, Klingon, and Dothraki get a second life?
It is the availability of the internet to the commoner that seems to have made all the difference. It is essential nowadays and I am not talking about your wifi and TV only – think smartphones, smart-wear, plus the Internet of Things (cars, refrigerators) spreading like wildfire – you get the picture.
Humans may have trouble coexisting with one another. We tend to get on each other’s nerves, be aggressive, point guns at our neighbors, and ultimately wage wars. But despite what the Hollywood movies have taught us over the years in the man vs machine narrative, there seems little potential that computers will turn against people. They are generally peaceful unless used by a sick mind… that of a human being.
Hopefully one day, humanity will adopt the language of good? Toki Pona7, that is.