Learn a Language, Work for Free: Part 1

by Translation Guy on January 25, 2012

“Luis Von Ahm claims he can translate Wikipedia – all 2 billion words of it – from English into Spanish in just 80 hours. What’s more, he will not have to pay anyone to do the work.”

Von Ahn, Carnegie Mellon computer scientist and serial internet entrepreneur, plans to unleash hordes of language students, armed only with online language instruction software, to translate the Web. The plan is to turn the homework generated by students of these free online language lessons into professional quality translation.

So, learn a language while you translate. Or is it the other way around? What’s the difference? Von Ahn gets it both ways. “The crazy thing about this method is that it works,” he says.

Well, he would say that, wouldn’t he? And Von Ahn gets to say it at TED, because his “Ideas are Worth Spreading,” because the internet genius has done the same kind of thing already. Von Ahn is the brains behing reCAPTCHA, that awful little Turing test that forces you to prove your humanity every time you fill out a web form. CAPTCHAs are used by many web sites to prevent abuse by “bots” busy spamming everything they can. They are the child-proof caps of the Internet. But reCAPTCHA has a twist.

Look familiar? You’ve got to re retype the two words into the form to prove to the Web that you are not a bot. So the script to be copied has to be hard enough so that the spam bots can’t read it, but easy enough for humans to follow.

Globally, we do about 200 million every 24 hours, for a daily total of around 150,000 hours sliced from the  life of humanity. Von Ahn figured he would piggy-back a transcription service for this necessary security task by doubling everyone’s work load. In Von Ahn’s version, the second word is a reject from an optical character scanner. You key in the correct spelling of the hard-to-scan word and Von Ahn gets the money.  Since he doesn’t pay his transcribers, he can undercut transcription services whom trapped in the past, are engaged in the archaic process of actually paying their workers. Which is a good thing, according to Von Ahn, because now transcription is cheaper, so more can be done (for free),

Well, he would say that, too. It’s revealing of the age that such a life-sucking macro-parasite is lauded for his prowess in getting people to do something for nothing.

Duolingo, his new translation service, is now open for testing, although you’ll have to get on a waiting list (200,000 names so far) to join up, since the beta is oversubscribed. Lessons are available in Spanish or German and part of the instructional method includes translating written sentences and rating the accuracy of translations made by others. Multiple review by multiple students keep student errors from leaking through to the client side, in theory. Early reports are that the course work is fun and engaging, which will be absolutely essential to keep students coming back to translate for free.

From the student side, early signs point to great success. Whatever pedagogical friction caused by  the need for random acts of  translation, will surely be outweighed for users by the siren call of the Net, “free”. And at least people will be learning a language while doing Duolingo rather than simply contributing the involuntary servitude required of reCAPTCHA. But what about the translations? Speculation on that next time.


  1. Tolken says:

    Interesting I learned a lot from this article. Thank you. For my part I have been struck the past couple of days by what babelverse is trying to do. It was sold in as something that would be great especially for scarce languages. People all over the globe could get hold of an interpreter with their language combination. Interpreters or bilinguals would sign up to a community. Needless to say there was no renumeration involved. Now, the latest development of this has a nasty after taste. For the state of the union speech they invited interpreters/bilinguals all over the world to interpret the speech – for free of course. This seems like an attempt to crowdsource interpreting as well. How long before e.g. news desks use that type of interpreting instead of their in house interpreters?

  2. ‘Serial internet entrepreneur’ means this guy will be on to his next project soon (in all probability). I wouldn’t worry about this as a threat to more traditional business model translation services.

  3. Arlene says:

    I realize that most of the internet is in english and that many languages do not have a ton of resources, but is translating all of it a bit much? Does he plan on translating Wikipedia into any other languages?

  4. I guess you get what you pay for. All for cheaper things (including translation work) but if I haven nott paid for it I have no right to complain if the translation isn’t what I expected.

  5. Come on, Ken, tell us how you really feel.

    • Ken says:

      I’m just getting started, Melinda…stay tuned.

  6. I read that one twice. A lot of info… and I finally know what those silly written words are for! I have been wondering that for a while. Thanks for the information.

  7. If Duolingo allows for the learning of languages at the same time, sounds pretty good. Not sure I want to be on a list of 200,000, but I may just have to check it out in the future when things settle down. I would be interested in looking at the quality of the lessons.

  8. Tracy says:

    Von Ahm says he doesn’t have to pay anyone, but does he get paid? Obviously he gets something out of this. I’m not sure about having boatloads of university students doing free translation for a grade. I know that I didn’t always put in the best effort for my grades.

  9. Gregor says:

    I urge everyone to watch the TED speech. While a professional translator could easily poke holes in his pitch, what I found creepy was the audience response. Based on the applause/laughter, his listeners seem to think getting cheap or free labor out of lots of people is cool. I suppose that’s because they see themselves as the exploiters, not the exploited.

    I was also flabbergasted to learn, in the course of researching Von Ahn, that my struggles with reCAPTCHA are not accidental. I had noticed lately that the CAPTCHA images were getting harder and harder to read. Turns out it’s because those twisted little images are now being lifted from digitized books, and by decoding them, I am helping someone/somewhere bring old printed books online. Had I been informed of this fact, I probably would have gone along with the venture. But I’m enraged to learn that in the course of trying to email a New York Times article to a friend, by solving a reCAPTCHA image I’ve been used for something else. Even if the ultimate goal is good, it’s dishonest to trick people and secretly divert their labor for another purpose.

    • Ken says:

      I share your rage, Gregor. Stay tuned for Part 3 where I show you how to get even. Also, my house is a block from Penn Station, so let me know next time you are in the city so we can have a drink.

LiveZilla Live Chat Software