06 70 33 24 905
News
Chasing the Unicorn of Auto-Translation: Why Automation Tools Cannot Work for Businesses
2014.08.22

This past month has seen some explosive growth in the world of auto-translation tools. It all started with Microsoft’s “Star Trek-like” innovation for Skype. By using a combination of Bing Translate and voice recognition, the company created a real-time voice translation service. Next, Twitter unveiled a translation service, so fans of soccer (or football, whatever you prefer) could tweet back-and-forth to each other, engaging around the globe as the games in Brazil unfold.
 
Image courtesy World Lens via YouTube.

Image courtesy World Lens via YouTube.

Another interesting development was Google’s acquisition of the app, World Lens. By offering a way to visually translate whatever you can see through your phone, this app offers users the chance to have a translator in their pocket whenever they need to read signs, menus or instructions. But does it work?

Kind of… as long as users don’t mind estimating the exact meaning of said signs and instructions. One user on Tech Republic said that a Russian sign that read “Parking Service Vehicles” was translated to “Stand Official Motor Transport,” rather than the seamless transition advertised by the app’s video.

I experimented with World Lens on some technical documentation and found about what I expected – an ingenious app that is great for consumers looking to just get by, but a mistake for businesses that think translation can be fully automated. Unless you don’t mind hacking through Spanish-to-English instructions like “Take to end the following procedure.”

The Legend of Automatic Translation

Google Translate helped make the idea of automatic translation “mainstream,” and the company’s most recent inventions make it seem like such a solution really is just around the corner. After all, if we can build self-driving cars, how come we can’t build software that eliminates language barriers completely?

The reason is simple: “language” is far more than just words. Take a look an encyclopedia to see how far words have evolved and how many different layers of meaning the words have. For machines to accurately translate, there needs to be as much abstract thinking as mathematical calculations. So until we can build a robot that is as artificially intelligent as a human, true translation automation may be a long ways off.

While that doesn’t make the efforts of Twitter (powered by Bing Translate), Microsoft or Google any less impressive, it does make it important for businesses to think carefully when planning global initiatives. The onslaught of recent innovations makes it tempting to look for automatic solutions to complex problems. Yet there’s an important difference between what consumers expect from free tools and what they expect from products.

Consumer Expectations for Apps v. Brands

YouTube once was a bastion of free-ranging, homemade videos. Anything could go viral, from a baby biting somebody to a group of puppies running down a hallway. Most of these clips were taken by amateur filmmakers, and the quality was poor, but authentic.

That has changed dramatically. Compared to the top YouTube videos of 2012, the top videos of 2013 were produced by professional teams and created by companies, rather than individuals. YouTube has become a mainstream destination; consequently, the most popular videos are more often professional productions than backyard iPhone videos.

Translation technology fits into this intersection of homemade and professional nicely. An app like World Lens may very well help you read a sign when you’re backpacking in Europe or choosing a sandwich in a restaurant in Mexico, but it’s not going to teach you how to assemble manufacturing equipment or manage a dashboard for cloud software.

The same goes for Google and Bing Translate and the other services it powers for the likes of Skype and Twitter. This was made painfully apparent during the big debut of Skype’s real-time translation feature at the Code Conference in Rancho Palos Verdes, California. Gurdeep Pall, vice president of Skype, had a brief conversation with a manager in Germany. The manager’s voice was translated to English instantly, but not without a few speed bumps.

“So what brings you to the United States, in addition to, of course, helping me with this demo?” Pall asked.

“I have many meetings with my colleagues in Redmond, and I take the opportunity to see her fiancé my,” the manager replied.

And, after a brief pause, according to Forbes, Pall exclaimed, “That’s nice!”

This kind of confusion is fine for casual conversations, but for business meetings, it’s sure to get frustrating fast. Likewise, Twitter, using the same technology, often confuses the order of words when translating tweets from another language into English.

That doesn’t mean translation technology can’t be used alongside human translators. In fact, many global language service providers do just that by using translation memory (TM). After going through collateral and locating key phrases and industry vocabulary, translators can save them to a database and have them automatically replaced across all the collateral using TM. That way, you have the oversight of someone who is familiar with the culture and context of the audience, empowered with—and still reaping the benefits of—automation technology.

The Cultural Divide

What automation really doesn’t take into account is the cultural divide between different markets. The graphical theme for a brochure in the US, for example, may not make sense when translated to German. Likewise, the tone of marketing has to be tailored for different audiences – while the US has a more casual tone in collateral, other markets are more formal.

So, while apps and automation technologies might work for consumers looking for a quick fix, they’re a long way off from working for businesses. Even in the same language, marketing and sales messaging has to be custom-tailored for each audience. Relying only on machine translation to convert those nuances to a new market could be a critical mistake. Culture is one thing that automated translated technologies still can’t detect. And as long as there’s no way to automate culture, there’s not going to be a way to truly automate language.

(Wired)

Languages

Hungarian, English, German, Russian, French, Portuguese, Spanish, Swedish, Italian, Czech, Serbian, Danish, Bulgarian, Croatian, Slovakian, Polish, Romanian, Slovenian, Flemish, Belorussian, Catalan, Dutch, Turkish, Albanian, Ukrainian, Greek, Bosnian, Catalan, Estonian, Finnish, Latvian, Lithuanian, Irish, Maltese, Armenian, Arabic, Hebrew, Thai, Japanese, Chinese, Korean, Vietnamese

1x1 Translations Ltd.

1DayTranslation.com

Phone: +36 70 33 24 905

Email: info@1daytranslation.com

Skype: onebyonetranslation

All rights reserved | 1x1 Translations ©
Website made by: