Duolingo's crowd-sourced translation model offers the best translation, considering people will interpret a sentence differently. At the same time, these repeated translations give Duolingo the capability to provide feedback on translations, as well as weed out troublemakers, the company says.

Other users will provide feedback too, and vote on the best translations. Duolingo feels this additional human feedback enhances the comprehension and learning process, and provides valuable perspective since often there's more than one correct way to translate a word, phrase, or sentence.

"When we started working on Duolingo we noticed that most language-learning methods outside of the classroom didn't do this," co-founder Luis von Ahn said of the system. He sees the feedback system as essential for learning and believes it is what keeps people coming back.

If the site's early success is any indication, von Ahn is right. The private beta attracted some 125,000 users to Duolingo, which translated over 70 million sentences. 30,000 of these users spend at least 30 minutes a week on the service, the company says.

For more tech news and commentary, follow Ed on Twitter at , on , or .