The startup Duolingo has found a way to make free foreign language learning profitable.
Venir otra vez?
The company, which counts Ashton Kutcher and four-hour obsessive Tim Ferriss among its backers, is doing it, in part, by enlisting new language learners to translate the web.
It's an approach that investors clearly like: The Pittsburgh-based startup today announced a new $15 million funding round led by NEA and Union Square Ventures. Duolingo launched in June with $3.3 million in backing from actor Kutcher's A-Grade Investments, Union Square Ventures, and Ferriss.
Duolingo users can currently learn Spanish, French, German, or (primarily in foreign markets) English through the company's sites. Activities are a mix of Rosetta Stone-style writing, speaking, and photo identification practice sets mixed with translation exercises. In the translation exercises, users parse level- and theme-appropriate Creative Commons content from the web and translate sentences they can understand at their learning level. Users then vote on the most accurate and most well-written translation. Like many other online learning sites, Duolingo has embraced the gamification model of language learning—its students are rewarded with points and badges for accomplishing specific tasks within their course.
Investors have given Duolingo more than $18 million primarily because of the site's profit model. Starting in October, CEO Luis von Ahn told Fast Company that Duolingo would begin accepting uploaded documents and web pages to be translated from the site's user community. In the near future, Duolingo will also begin to accept paid translations. In exchange for payment, Duolingo will steer users toward the translation of premium documents. The company's business model relies on the idea of offering accurate crowdsourced translations by motivated amateurs to the larger web.
Before launching Duolingo, von Ahn was best known for helping to develop the ubiquitous CAPTCHA tests found online and for developing reCAPTCHA in 2007. ReCAPTCHA was a predecessor to Duolingo in terms of crowdsourcing; over 40 million words a day from books in the process of digitization are transcribed by web users thanks to the service.
Duolingo sorts content by algorithm. Users just starting to learn a language are steered toward simple content with easy-to-learn words such as Wikipedia pages on athletes and video games; as users progress, they begin to translate content with more complicated grammar and vocabulary. Translation tasks are keyed to learning exercises. Language learners in the middle of the unit on food are likely to translate recipes, while someone in a grammar unit is more likely to translate a travelogue with multiple tenses.
While language learning is big business in the United States, Duolingo is looking at the international market for user retention. As von Ahn notes, American language learners tend to be hobbyists, while foreign English-learners are primarily interested in language learning to improve their career prospects. Meanwhile, Rival Rosetta Stone is currently in talks with Groupon for an undisclosed international expansion project.
[Image: Flickr user INDIVIDUELL MÄNNISKOHJÄLP]