The End of Translation as We Know It?

The End of Translation as We Know It?

The End of Translation as We Know It?

The End of Translation as We Know It?

Professional translators take pride in their linguistic expertise and ability to match a certain tone or voice in their target country. They know the culture, the language, political situation, what’s in or out, and how to address different audiences like developers, administrators, new users, and teenagers. But with all the hype about NMT, combined with perpetual cost and deadline pressures, is the profession of the traditional translator dying?

The world is changing

And so do translators. They must adapt to the changing way we communicate through different channels on different media and platforms, because the content that must be localized is always evolving. It has become more versatile. Information provided on webpages demands a different approach than software strings and dialogs. Some content has a very short lifespan and needs to be turned around rapidly (GALA recently discussed this in their presentation, Continuous Delivery in Localization—Achievable Reality?).

The tools industry has not left translators to face these challenges alone. MT has been substantially enhanced with the implementation of neural networks. The fluency of content localized through NMT has vastly improved, although now we must deal with simple word-based errors that are not easy to spot if under time pressures (hear more in our podcast, The Challenge of Neural MT: Part 3). CAT tools are combining MT with recycling, terminology, and user entry prediction (e.g. Lilt). Some even provide speech entry functionality. All these features can become productivity boosters, but require a totally different approach by the translator.

What is the future of the translator?

The translator’s job profile will become even more diverse than it is today. A huge trend to meet the challenges of rapid turnaround and decreasing prices for commodity localization is machine translation post-editing (MTPE). This task, however, is not everyone’s cup of tea. Not only because it requires a different approach and skills than localization from scratch, but because post-editors also have to adapt to the different qualities of the automated localization, be it recycled from TMs or from the various incarnations of MT, such as statistical, phrase-based, rule-based, and neural (see our recent podcast series on neural MT).

In many areas, technical quality predominates linguistic quality. The MT post-editing process is taken on not by translators, but by subject matter experts who do not have to be bilingual. As experts in their specific area, these editors are not clinging to the source content, but rather are correcting clumsy machine translations with a focus on technical correctness. (Go here for some research on this topic, and here for one example of a major industry leader adopting MT and PE.)

Brainpower still in demand

Without a doubt, there will still be a need for language professionals. Translation of technical documentation such as Life Sciences, legal content, as well as literature pose very specific challenges to translators and require linguistic excellence. In fact, seasoned translators with a lot of experience usually do not find MT helpful to their work.

Then there’s the area of marketing localization and transcreation. It’s vital that experts in this subject matter are located in the target market. They need to “speak” the language of marketing and advertising, be familiar with the local scene, and know what is currently trendy. To create relevant content, using the market’s specific terminology is key.

Another low-cost human translation technique becoming more popular due to continuous price pressure is crowdsourcing. Didn’t we shrug off that approach as unrealistic and think the quality would not be good enough? Think again. Considering the short life span of some content, and the decreasing quality expectations of end users, this solution is being reconsidered to get more bang for your buck.

So yes, the role of the translator is changing. New skills are required depending on the type of localization deliverables and on the specific process. As with everything in today’s fast-paced world, you can’t linger behind.