Duolingo Beta put to the test: can you learn a language while translating the web?

February 29th, 2012 by

A while ago, while trying to avoid some work, I watched this TED presentation by Luis von Ahn.

While his name might not ring a bell for everyone, his legacy definitely will: he is the guy behind reCAPTCHA, the bot-blocking service that uses all of us to help digitize the archives of The New York Times and Google Books. (Just think: every time you have to squint your eyes and turn your head sideways to decipher those annoyingly blurry squiggly words, you’re making the world a better place. Still, feel free to swear while you do it.)

In this TED presentation, von Ahn introduces his new project, Duolingo. Duolingo’s goal, in short, is to crowdsource language teaching. Here’s the twist: users will be able to learn a new tongue free of charge while translating the web through (unpaid, obviously) microwork.

Given the revolutionary nature of this idea, you can understand my excitement when the invitation for the Duolingo Beta landed recently in my mailbox. (It was also a great chance to refresh my rusty Spanish skills, recently put to the test – with mediocre results – while exploring a notorious Caribbean island.)

A brilliant user interface

At first glance, Duolingo is clean, friendly and sleek. There is no hint of the annoying and unintentionally funny user interface associated with reCAPTCHA.
To begin with, the platform gave me a quick tour, introducing really basic concepts and allowing me to try my hand at translating simple sentences. At any time I could hover on each linguistic particle to see a suggested translation.

I was then introduced to a skill tree, which some gamers out there will be used to, but is novel in a language teaching context.

Upon starting the first “lesson” (“exercises” would be a better term for what they are in practice), three familiar hearts appeared on the top right of the window. This is a way to check quality: if you lose all three you must start over.

After a few excercises I was humbled by my very first achievement. I’d like to thank the Academy (and my parents) for making this possible.

Then it was time to do some actual (micro)work, translating the web. At this point the interface changed slightly. I was asked to translate a sentence, which could be played in audio format and had an accompanying picture to help me understand the concept in context.

The actual sentence was “Ilumina la piscina”. I went with a straightforward (in my mind) “Light the pool”. It turns out Duolingo demands precision: two other users provided a better translation (“Illuminate the swimming pool”). I lost one heart. Ouch!

The second sentence went better, as my translation matched the others.

For the third sentence I scored a “60% agreement”, meaning that one user translated the sentence like me and the other suggested a second option.

After the check, Duolingo asked me to rate the other users’ performances. I assume this is their way to crosscheck the quality of translations. It would be interesting to know how many times they need to do that before being sure of an answer (industry standard, as far as I know, is around 7 times).

…And the first task was complete. More achievements, points and unlocks! The gamer in me rejoiced.

My verdict

It’s easy to get excited about Duolingo (especially if you don’t get out much). Given it is estimated that only 0,5% of what needs to be translated today is actually translated (due to high cost and other contraints), if it takes off it could really change the world.

So should the rest of the translation industry be afraid? I decided to ask an expert in the matter, Jani Penttinen. Jani said he can’t see how Duolingo could sufficiently guarantee quality to make it a serious threat to professional translation services, but he does “see it as a great way to get translations to consumers, in areas where it is not currently possible to translate content due to high cost.”

In my view, Von Ahn and friends have done a great job of designing the interaction between users and the platform so that the concept works as intended. That alone is a huge achievement. I would need to spend more quality time with Duolingo to see if it will really help me learn Spanish, but it certainly is much more appealing than boring grammar books. Now, if they could only add Finnish to the list of available languages and help fix that little drama of mine…

  • Jaimito Dc

    Thanks a lot for this analysis.
    A couple of questions:Is this the whole of the user experience?Is there a place where I can actually learn and practice vocabulary?
    What type of interaction can I have with other users?

    I’m interested to see how I actually learn, does the skills tree teach me grammar or vocabulary?
    Thanks again, 

  • http://www.parliamodivideogiochi.it Tommaso De Benetti

     Hi Jaimito. So:
    1. No, this is not the whole user experience, there are other tasks to be performed (pronunciation for example)

    2. There are ways for you to learn vocaboulary in context, during translation. There is a clear separation between “lessons” and straight translation from the web.

    3. Other users: as far as I saw (didn’t continued too much after writing the article because I have more pressing things to take care of, but I definitely will use it more in the future), you can’t interact with other users directly, just rate their performances. Take this as a preliminary impression, things might change down the skill tree.

    4. See above. I think the approach of Duolingo is teaching you grammar “in context”. I kind of like it, it’s way less borning than a school book.