Tuesday, May 03, 2011

Duolingo: Crowdsourcing at its Best for the Translation Industry

For the last month I have been reading tweets and notes about Duolingo as the place where "you learn a language and simultaneously translate the Web," but I kept postponing getting more information about it. As a good procrastinator, I figured that if this was really important, it would eventually make its way to me. Well... it did!

Ultan Ó Broin mentioned my name in his Blogos entry "The Future of Web Translation: Haters Gonna Hate" and I felt compelled to watch the video by Luis von Ahn, the inventor of Recaptcha, at a recent TEDx event at Carnegie Mellon University.

Duolingo does for translation what Flickr did for photography and what Wikipedia did for encyclopedias. It brings the knowledge of amateurs to do some work that only professionals could do. The advantage of Duolingo is that − unlike Facebook or Hootsuite, who also use community translation − the user learns a language in the process.

Crowdsourcing − the approach used by Duolingo − is the act of outsourcing tasks, traditionally performed by an employee or contractor, to an undefined, large group of people or community (a "crowd"), through an open call. The goal for Duolingo is to get 100 million people to translate the web into every major language for free.

The trade-off here, according to Luis, is that there are 1.2 billion people in the world learning a second language and they have to pay for it. With Duolingo, they will learn a language for free and translate the web in return. A really revolutionary and innovative concept.


According to Luis, using this approach, Wikipedia could be translated into Spanish in five weeks with 100,000 people or in 80 hours with one million individuals.

Who does this approach benefit? Everybody.

Who does it hurt?
  • Insecure translators who like to complain about things they can't control.
  • Rosetta Stone, Livemocha, Fluenz and other software-based language learning software.
  • Machine translation providers like AsiaOnline, PROMT and SDL, because Duolingo could be a faster/better solution.
Duolingo is a welcome addition to the arsenal of language solutions around the world. It is clearly a solution for making knowledge and information that would never be professionally translated available, especially in languages where the translator pool is insufficient for the amount of content that is available for translation. Watch out Google Translate!

18 comments:

  1. Renato

    Having been involved with translating the Wikipedia, I can assure you that language students can not translate the wikipedia in "80 hours with one million individuals" simply becuase there are too many highly specialized terms that need real subject matter expertise (SME) and real linguistic expertise to resolve.

    For some areas you need engineering students not language learners.

    However, I agree this is a very cool concept but I think the more likely scenario is that MT, Crowdsouring and forward thinking Professionals will create new production eco-systems that get large amounts of high value content translated at relatively high quality for relatively low cost.

    And there is nothing stopping all the language learning companies from also extending their businesses to include MT, and professional translation process management to also be players here.

    ReplyDelete
  2. I am just copying some comments in from LinkedIn

    It is hard to tell at this point with the scarce details, but I think as an impartial observer I can say the following:

    * This is not likely to be a direct replacement for professional translation services as there are too many loose ends here that need more definition

    * This looks very plausible for all kinds of volatile content, especially social media content

    * If this works it could affect at least some of the low-value work done by professionals

    * While it seem unlikely that a complete novice in the target language can perform any kind of useful translation, it is quite possible that an advanced beginner can begin to contribute and that intermediate student can do simple sentences.

    You can already see the professionals bashing on the Duolingo concept in Twitter. e.g. In Twitter you can see messages like: “Can someone explain how you can learn Japanese from scratch by translating it? #WTF #nothatingjustasking”

    This, by the way, is also a concept that any language teaching business can employ in some way to break translations into small chunks and use it to teach students and reduce their fees or even actually pay them very low rates to do homework that can be charged to a customer. So this is a just a new model linking language learning to translation and nobody says that the only possible way to do this is to get people to do it for free. Pay models are possible.

    I am sure there will be other approaches that can mix free with pay to get “functional bilinguals” involved in doing translation at much lower rates than the industry would like to see and again many of the people who decide to engage in this may use MT in combination to speed up their translations knowing quite well that MT is imperfect and needs to be corrected.

    Collaboration infrastructure (software) that facilitates and manages this flow can make much of this easier and as tools like the Microsoft CTF already show, these tools are already integrating MT, TM and open volunteer collaboration together for motivated communities to translate content they feel is valuable.These kind of tools I think will eventually make TMS system much less relevant if not obsolete.

    Duolingo will very likely stir up a lot of hate as it is going to affect at least some (who add very little value) in professional translation.

    It is important to realize that the motivation behind this is not to marginalize LSPs and "localization professionals", rather, it is a driving human need to get access to knowledge and information that is locked within a few languages and make huge amounts of useful content available to more people in the world. To my mind this initiative is just a step along the way to a world where 100X to 1000X more content will be translated.

    Together with MT, such kinds of collaboration initiatives are absolutely necessary to make this kind of translation volume possible.

    In many ways the only thing new about this is the potential scale which is yet to be proven but seems highly likely given von Ahn's track record.

    ReplyDelete
  3. Joan Parra6:16 PM

    I learned about doulingo for the first time in what I thought to be a tongue-in-cheek post from an usually knowledgeable and sensible source. But hey, now I see it's serious. Well I'd really love to learn Russian (or Hungarian, for that matter) for free and directly from my desk so please would you be so nice to let me know when I can start translating from it. At no charge of course.

    ReplyDelete
  4. Joan, If you just want to learn the language for free you can go to www.livemocha.com and they have basic courses in Hungarian and Russian there. When I was stuck in Stockholm during the volcano airport shutdown last year, I took a few classes in Swedish and I might be able to remember a few words if I try hard enough.

    I don't know how Duolingo will work, but there are many ways in which I can get a person to translate concepts without knowing the other language. If you show me a picture of a cup, I will easily identify it as a "xícara" ou "taça" in Portuguese even if the original caption is in Korean.

    I have taught languages in the beginning of my career and I was always amazed at the pace of language acquisition for people who are motivated. I am sincerely looking forward to the process that they have identified to make this feasible.

    Let me put it this way: I choose to believe.

    ReplyDelete
  5. Joan Parra7:30 PM

    Thanks for your time and attention, Renato. I still have some doubts though. For instance, why would anyone choose to pay (with their work) for something they can get at no cost at livemocha e.a.? By the way, even chimpanzees are able to identify a cup when they are presented with it. However, i wouldn't request a translation from a monkey even if it was for free.

    ReplyDelete
  6. Joan, you would be surprised at what people do for free. In any case, if you watched the video (if you haven't I strongly suggest you do), you will see that your valid concerns are addressed by the presenter. Including the fact that many people don't know that when they use Recaptcha to sign into a website, they are actually working for free for the New York Times.

    ReplyDelete
  7. Joan Parra8:36 AM

    I'm aware of that, Renato, and I have seen the video. I know lots of people are willing to work for free for a number of reasons and I don't see that as a threat for my business. I don't own an LSP, I don’t work on web localization nor do I have any vested interest.
    I just find the idea behind duolingo very questionable.
    1. Having people translate from languages, cultures and subject matters they don't know IS foolish. First, it implies that translation is an empty, brainless activity, void of any complexity, which ANY individual is able to perform. Well I don't agree with that, also for a number of reasons, none of which has to do with hate, insecurity or low professional profile on my side. I'm not academically nor professionally biased, I just happen to have a certain amount of education and experience that prevents me from sharing the over-simplistic half-baked views of some nerdy entrepreneur.
    2. Learning a language outside a real-life communication environment and without a reasonable progression scheme is learning the hard way, at the very least. Using translation as the main methodological approach doesn’t make it better. It's a lame idea that gets us back to the old days when schoolboys "learnt" latin and greek by translating Caesar and Homer. For the record: It DIDN'T work.
    Disguising that poor scheme as a 'fair business model' doesn't change those simple facts.

    ReplyDelete
    Replies
    1. Joan, I am a translator and I think exactly the same as you. I am really upset when I hear things similar to what Kirty said that is, that Duolingo will only "affect at least some (who add very little value) in professional translation".

      That assumption that anyone who complains is a "bad translator" is really frustrating, biased, generalistic and terribly simplistic.

      I have pointed out Duolingo's possible problems and its basis and some have called me "condescending", "egotistical" for "underestimating" learners.

      Instead of questioning and being critical about what "seems" to be a good idea, they only praise the idea and classify those who don't like Duolingo as "bad translator". Under what basis are these people judging me and everyone who questions Duolingo if they do not know nothing about us?

      It's sad, because instead of looking at the good side and bad side of Duolingo, in other words, to what really is, many prefer to put the "good side" as if the "bad side" did not exist.

      One has studied lots of years, translated many texts, and suddenly, you are classified as an "ignorant", only because you don't agree on something that one truly believes it won't work. It really is a shame.

      Delete
  8. Good idea, great presentation. If he avoided the unrealistic buzzword statements like "We believe language education should be free", "It doesn't discriminate against poor people", I would give him a 10 for this presenation.

    This will work. Even if I only think about their approach to reCaptcha and combine it with Rosetta Stone technique, it is easy to see the potential. And I assume he would not do this presentation if the translation accuracy was lower than current MT stats on average.

    Jean, it is hard to disagree with your points. My perspective on this is a bit different and I do not consider this as a replacement for neither commercial translation nor commercial language education. Open source software has never been a replacement for professional software development. Projects that attract masses (like Firefox) succeed immensely and do kill the commercial options. However, while doing so, they expand the space and open more room for new and larger commercial offerings. With similar logic, this will work well for Wikipedia but perhaps not as well for a technical manual about a rocket. It will never replace the experience of taking an ESL class in an English speaking country.

    I do own an LSP but like crowdsourcing, with its current state, this has little to no impact on a professional LSP with (real) knowledge of the industry.

    ReplyDelete
  9. By the way... Renato, it is probably time to switch your blog to reCaptcha, I just wasted 10 seconds... :)

    ReplyDelete
  10. Professional translators --much to the regret of Renato and others-- charge, sometimes high fees, for their work because translation implies an intellectual effort --and quite an effort, sometimes.

    For a crowdsourcing approach to translation like Duolingo to be valid, its outputs must remain free, that is, people should be granted access to them and be entitled to use them without having to pay any fees or buying any licenses whatsoever. That is the approach followed by really fair crowdsourced solutions such as Wikipedia. I can't imagine Jimmy Wales asking people to contribute with an article --or a piece of an article, for that matter-- each time they visit his website and then trying to sell the resulting work to someone else for a profit. Duolingo is actually asking their users to contribute with their intellectual effort each time they visit the site to practice their language skills. And the promise of learning another language remains just as that, a promise. It is entirely up to the user to achieve his or her language learning goals, very much like any other approach to language teaching. Oh, not to mention that Luis talked about "monetizing" the website by selling the translated content to others.

    You are right, Renato I am surprised at people's enthusiasm about Duolingo. It is much like an "internship" approach to translation. Just for the fun of it, let's do some intellectual effort --easy, I am not going to charge you for this-- and let's extrapolate the business model to a different field. Let's say, software programming. It means that a wannabe programmer would visit a site, say, Duohacky, to learn how to program in C++. There are many websites that offer free programming classes, but this is different in that you learn by doing. And not only that, it is different in that you will play with real life lines of code! So, the idea is that he or she learns how to program in C++ by contributing with some lines of code for a given software development project each time they want to practice their programming skills. At the beginning, it will be easy stuff --we don't want to scare away our beloved learners-contributors-- and as you gain experience, you will find more difficult stuff. Well, that's it. Voila! Now the owner of Duohacky can complete a $20 million software development project in a blink just by using these wannabe programmers without paying them a single cent! Ha! And they thought they were getting something in exchange!

    With such an smartass business model, it's no wonder that programmers hate it... Oh, probably those who hated it would be insecure programmers who like to complain about things they can't control! Well, back to the translation arena, who would trust a $50 million translation project to a bunch of wannabes/ interns? Oh, I forgot they wouldn't have to pay $50 million. Mmm... but this is supposed to be run as a business, so they would have to pay anyways. Let's charge them, say, $500.000 instead. That can't be bad! They are saving tons of money! And they wouldn't even notice it is done by amateurs! Oh, and don't forget to include a BIG clause in the client's contract saying that the "translation product" is delivered "as is" and that the Company is not liable for any mistakes and bla bla bla... "Dear client, you blame the crowd. They are error-prone, not us." You, entrepreneurs, make me sick.

    (To be continued...)

    ReplyDelete
  11. Translation was, has been, is and will always be a profession for one single reason: it takes years of study and practice --meaning a huge investment in time and money-- to create a reliable translator. And the majority of content that is worth translating needs people who is able and willing to be held liable for every comma and every character they put in a translation. I understand that lots of people would like to get professional translators out of the equation, particularly those who are feeling our impact on their margins. To my understanding, that is the only reason some fat businessmen are so eagerly advocating solutions (TMs, MTs, amateur-based crowdsourcing) that pretend to replace professional translators. Even though every person having a working knowledge of two or more languages is capable of translating short simple sentences --much like anyone can add two and two--, any serious attempt at translation requires --at some point in the production chain-- the services of a language professional or a subject-matter professional with excellent language skills. Otherwise, that translation attempt is at risk of falling in disgrace. The Web is plenty of examples of "bad" translations, whose existence is only explained by the fact that the role of the translator was mocked by someone lacking the required skills.

    By the way, I am not against crowdsourcing nor against people learning languages, not at all. Neither am I scared about being forced out of business. I really don't think that will happen. And if it does, I can always reinvent myself. But I will fight every attempt by others to minimize or ridicule translators' role in society. It is --and it has always been-- us who are really interested in putting culture and knowledge in the hands of the common people. People like you, instead, are only interested in making people believe that a botched translation is worth buying.

    Finally, just to make it crystal clear, Renato: whatever approach to crowdsourcing that implies someone profiting from the crowdsourcers' intellectual efforts is UNFAIR. It is easy to pay lip-service to an effort when you have nothing to lose and much to gain. In other words, as a friend of mine said, it is easy "to beat the nettle with someone else's dick".  At least, those crowdsourced ventures, such as ReCaptcha, should have an opt-out or --even better-- an opt-in button to be deemed fair. Yes, I am surprised at what people --unknowingly-- do for free. But I am thrilled at how people reacts when they learn that they have been doing so. You don't believe me? See for yourself: karouselmag.com/2010/10/recaptcha-exploits-the-masses/

    Regards

    ReplyDelete
  12. I don't know. As an advanced speaker of Japanese who has worked as a freelance translator (JP>EN), I find the idea of Duolingo interesting. It has the potential to give me the opportunity to expand my knowledge of Japanese into more specialized topics while contributing to the community at large. As something to do in my free time, it is good practice (and something to add to the old CV) without the time constraints of a contract translation job.
    That being said, I have yet to learn more about Duolingo, so maybe it doesn't work the way I think it might.

    ReplyDelete
  13. (Continued...)

    Translation was, has been, is and will always be a profession for one single reason: it takes years of study and practice --meaning a huge investment in time and money-- to create a reliable translator. And time and time again you find how difficult it is to get rid of professional translators. Besides, the majority of content that is worth translating needs people who are able and willing to be held liable for every comma and every character they put in a translation. I understand that lots of people would like to get professional translators out of the equation, particularly those who are feeling our impact on their margins. To my understanding, that is the only reason some fat businessmen are so eagerly advocating solutions (TMs, MTs, amateur-based crowdsourcing) that pretend to replace professional translators. Even though every person having a working knowledge of two or more languages is capable of translating short simple sentences --much like anyone can add two and two--, any serious attempt at translation requires --at some point in the production chain-- the services of a language professional or a subject-matter professional with excellent language skills. Otherwise, that translation attempt is at risk of falling in disgrace. The Web is plenty of examples of "bad" translations, whose existence is only explained by the fact that the role of the translator was mocked by someone lacking the required skills.

    By the way, I am not against crowdsourcing nor against people learning languages, not at all. Neither am I scared about being forced out of business. I really don't think that will happen. And if it does, I can always reinvent myself. But I will fight every attempt by others to minimize or ridicule translators' role in society. It is --and it has always been-- us who are really interested in putting culture and knowledge in the hands of the common people. People like you, instead, are only interested in making people believe that a botched translation is worth buying.

    Finally, just to make it crystal clear, Renato: whatever approach to crowdsourcing that implies someone profiting from the crowdsourcers' intellectual efforts is UNFAIR. It is easy to pay lip-service to an effort when you have nothing to lose and much to gain. In other words, as a friend of mine said, it is easy "to beat the nettle with someone else's dick".  At least, those crowdsourced ventures, such as ReCaptcha, should have an opt-out or --even better-- an opt-in button to be deemed fair. Yes, I am surprised at what people --unknowingly-- do for free. But I am thrilled at how people reacts when they learn that they have been doing so. You don't believe me? See for yourself: http://karouselmag.com/2010/10/recaptcha-exploits-the-masses/

    Regards

    ReplyDelete
  14. Renato, I'm not sure about this approach. I don't think it will hurt professional translators, since anyone who wants quality knows that s/he should hire a professional. I am actually concerned about the kids: What kind of information will be available to them? Where's quality? I have to constantly tell my daughters that stuff they hear in American TV shows dubbed into Portuguese, for instance, is wrong. It's really, really poor Portuguese. And sometimes plain wrong. And these translations are usually done by "professionals" (auditions become audições, for instance - I can't remember any other examples right now, but there are plenty, believe me). So I wonder what kind of stuff my kids will be learning on the internet by reading texts translated by language learners. I fear for the future!

    ReplyDelete
  15. I got very alarmed and interested at the same time about this idea. Thank you for sharing this. As a professional translator specialized in Korean translation, this is a matter I cannot let go unnoticed. Koreans are, I'm quite confident, more than willing to participate duolingo, if it offers what it says it does. I was a little skeptical about the idea of crowd-sourcing when I first heard it, but the concept of duolingo now makes it more plausible that professional translation is facing something big. Thank you.

    ReplyDelete
  16. Interesting and nice design, but I prefer "Studystream" by FAR!
    WOW! That's impressive, fun and I don't need to do boring translations to learn for free :P
    Check their video in NYC BigApps3.0: http://2011.nycbigapps.com/submissions/5813-studystream-online-language-learning

    ReplyDelete
  17. Anonymous10:58 AM

    I'm not a translator. I have a duolingo user.
    It's good. There are a decent range of lessons similar to what a tutor would go through in a class.
    - listen to this Spanish sentence and write it out
    - translate this sentence into English
    - translate this sentence into Spanish
    - repeat after me (it uses your microphone)
    - learn this word
    - choose the correct verb form

    For an early stage learner it's probably almost as good as a one on one lesson. You can do it little and often instead of in a chunk once or twice a week, give yourself time to mull things over, you need to be extremely precise or it marks you wrong.

    Then you practice by doing the translations. Practising on translations is something I find that works. I see new words, get familiar with the language and its structure. Instant feedback on accuracy is provided, other members can rate my translations as good or bad. The material tends to be more topical then anything I've seen in a text book, and therefore easier to engage with.

    It doesn't make translation into an amateur job, it doesn't make professional translators jobs seem useless. It actually highlights the skills involved. It takes hundreds of amateurs to do the job one skilled professional can do.

    Does it threaten translators. Probably but not to the extent people fear. Dense, highly technical stuff, the sort of material that needs to be translated accurately and precisely. That's the meat and veg of professional translating work. That's not what duolingo will translate. Duolingo is for the stuff that doesn't need to be 100% accurate, close enough will do. It's for the stuff people won't pay a professional translator to do, but might be prepared to pay someone a tiny fee. It might end up translating 1000 times more than professional translators do at the moment, but that will only be because the translation market will be 1001 times as big.

    and if it doesn't nail it as a language learning device, then it won't translate anything at all.

    ReplyDelete