I recently participated in the TAUS event Reinventing the Translation Industry.
The focus of the event was on imagining the translation industry a year from now, in light of our global economic and pandemic woes. But also with an eye on how machine translation (MT) and AI are transforming the industry to support more content, more languages and more intelligent (and efficient) workflows.
After a few days of presentations and discussion, I feel the general takeaway was that the year ahead is less about technological disruption and more about the execution of technologies that already exist. With budgets frozen (or cut) it’s safe to say we’re in an austerity phase across many industries, which will force companies to look for efficiencies wherever they can find it.
And machine translation and AI have important roles to play in this regard. Too many companies continue to hold MT at arm’s length instead of looking at the many ways it can unlock content and result in a more demand-driven translation workflow.
What follows are a few themes that stuck with me from the conference…
Removing humans from the MT workflow
With a handful of exceptions, MT has been used behind the scenes — as a first-pass translation tool with humans in the center of the workflow. But the emergence of neural machine translation (NMT) and the improved quality of output has given some organizations an opportunity to test MT without human intervention. Certainly, customers are already well accustomed to self-serve machine translation, warts and all. But companies have been understandably more cautious, though I believe the need to unlock content will ultimately pull more and more organizations into diving into the MT waters.
Wayne Bourland of Dell noted their work on rolling out MT. Currently, MT plays a role in roughly 70% of their content, typically with human involvement along the way. In five years they aim for only 10% of content requiring human touch; this is a bold goal, but one that I suspect they will achieve.
The vast majority of small and medium-sized companies have done little with MT so far — and, in my view, represent the greatest opportunity for vendors in the near future. These companies are eager to expand globally and can move quickly.
To be clear, I’m not suggesting that mission-critical content be left alone to MT. I am suggesting that organizations with millions of words of knowledge base content, tutorials, customer reviews, etc. — with no foreseeable budget for professional translation — will see the value of MT. Their customers most certainly will.
Are project managers most at risk in a more automated future?
A natural question to ask about this automated future is “who stands to lose their jobs?” I don’t think most professional translators are in danger here — though it was noted that translators should evolve into “digital linguists” to take full advantage of these emerging technologies. And I agree. Translators who are experts with MT technology are well positioned for a more automated future.
Translators also provide clients with a great deal more than simply translation — informal consulting on product names, images, functionality. And often these services are not monetized — and should be.
But what about project managers — the people who manage the translation workflow — within client companies and within language vendors? There are technologies evolving to automate a growing percentage of what these folks do today. So I would recommend that project managers, like translators, move up the “value chain” within their organizations to the additional services they can provide that computers cannot.
Is lack of language support a form of discrimination?
In my presentation, I drew from the Web Globalization Report Card, pointing out that the leading global companies support, on average, 33 languages. This breadth of language support reflects significant investment. When you step back and look at smaller, emerging global companies, the average number of languages drops to 10 or fewer.
Nevertheless, Wikipedia (which emerged as the best global website in the Report Card) supports 290 languages, which is a far better reflection of true global language demand. Clearly, vast numbers of Internet users are not well served linguistically.
When you don’t support someone’s language on your website, you send a message that they are not as important as other people. Now I realize this message is not intentional. There are only so many dollars available for professional translation.
How do we get from 33 languages to 290 languages? Without machine translation, we’re not going to get there quickly I’m afraid. That’s why MT services such as Google Translate (with support for 108 languages) are so critical to people around the world.
Also, I think it’s important for those of us in this industry to be more careful about how to speak of languages. For example, I still hear of languages referred to as “major” or “minor” languages based on the number of people who speak the language. I’ll confess, I’ve made this mistake in the past many times. But a language is not minor to the person who speaks it, regardless of the number of speakers.
So I was pleased during the TAUS event to see a proposed startup address “long tail” languages, which is a better description than “minority.” Perhaps by removing loaded adjectives such as major and minor, we can look at all languages more equally and, in doing so, look at the people who speak them more equally.
Translation is an act of optimism
I believe that we are in the optimism business. Organizations invest in translation because they are optimistic about growth and about providing a better customer experience.
These are depressing times to be sure. Which is why our industry has an important role to play in leading the way to a more optimistic future. And I felt that most folks at the TAUS conference are quite optimistic despite the challenges we face.
So those are my random thoughts from the event. I was glad to have participated and hope to again someday.