Apple's iPod Touch schematic.

Translating the Entire Internet, One Language Lesson At a Time


Photo: Karen Rubado / Flickr 

Luis Von Ahn wants to translate the entire internet. It may sound crazy, but his goal is for every website in the world to be translated into multiple languages, so that we can all access the internet equally.
Of course, he isn’t taking on this endeavor alone. With his company Duolingo, he’s enlisting the help of anyone who wants to learn a new language. Venture capital firms New Enterprise Associates (NEA) and Union Square Ventures are also backing him up with a $15 million investment.
On its face, Duolingo is a free site where fluent English speakers can learn French, Spanish, or German, and Spanish speakers can learn English or Portuguese. As you go through the series of exercises, you get sentences from the internet that you are asked to translate. In the beginning you’ll get simple sentences such as, “Me llamo Sarah” (Spanish for “my name is Sarah”). As you move up, you’ll be asked to translate more complex sentences or paragraphs. If you get stuck, Duolingo gives you suggestions for each word you’re struggling with and keeps a list of difficult words that it will help you memorize later.
Yet while you plug away at your language lessons, what you’re really doing is translating Creative Commons documents into a different language. Scattered throughout the lessons, Duolingo gives you one of these documents, broken down into sentences, and asks you to help translate. You can try your hand at as few or many sentences as you wish, and you earn points for each completed sentence. Duolingo will give you hints about what each individual word means, but it’s up to you piece the sentence together in a way that makes grammatical sense.  And don’t worry if you feel overwhelmed – your translation doesn’t need to be spot-on. The community of fellow students can help edit what you’ve translated.
Dr. von Ahn hails from Carniege Mellon University, where in 2005 as a graduate student, he created a game to label photos to make them easily searchable. Two years later, after Google acquired his technology and turned it into the now-defunct Google Image Labeler, Von Ahn built reCAPTCHA, a tool that websites use to verify you’re a real person. What many people don’t realize is that as they’re trying to decipher those twisted letters and numbers, they’re also helping to convertg scanned books into digital content. Two years later, Google bought reCAPTCHA, too.
Von Ahn’s fixation on leveraging the crowd to convert large volumes of content into something useful has carried into Duolingo, which launched to the public in June. While the company only uses Creative Commons licensed text for its lessons right now, the company has plans to offer paid translation services to website owners as a way to make money.
Of course, Google Translate already translates websites into 66 languages (many more than Duolingo can), but Von Ahn is bettingthat humans are better than computers at the job. Google Translate also doesn’t offer the kind of permanent solution Von Ahn is looking for, since it only translates a site for as long as you’re on it. But given Von Ahn’s success at selling companies to a certain search giant, the divide between Google and Duolingo could end up a distinction without a difference.

Leave a Reply