esterday I said goodbye to my colleagues from my interpreting traineeship at the Mediterranean Agronomic Institute of Zaragoza. I could write a dozen different posts about the lessons I learned over these three months (both literally and figuratively, since the material I interpreted consisted mostly of Masters lectures). Perhaps the other eleven will come later, but for…
Duolingo – harnessing human brain power for good
If you have an interest in languages, or spend much time in the App Store, there’s a good chance you’ll recognise this little guy: the friendly green owl of Duolingo. Duolingo is a platform that allows the user to learn another language in a fun environment and for free, which is a sweet enough deal to begin with. But Duolingo is more than that: it’s also an ambitious strategy aiming to facilitate the translation of the entire Internet – also for free.
Both of these aims are admirable, but when I watched this inspiring and entertaining TED talk about the creation of Duolingo by its founder, Guatamalan Luis von Ahn (talk available in Spanish here and in English here), I found that admirable doesn’t quite cover it. The video is 17 minutes long, and it’s well worth the time, but here’s a quick summary.
Before Duolingo, Luis von Ahn invented the CAPTCHA – you know, those annoying little sequences of letters you have to type out when you fill in a form on the internet. But before you start hating him for bringing those pesky little timewasters into the world, read on. The CAPTCHA asks you to complete a task that is simple for a human brain but impossible for a computer – to decipher a string of warped letters. Before long, 200 million of these fields were being filled out every day by humans across the globe – and if each one takes 10 seconds, that’s over 500,000 hours per day of human brain activity performing a task that computers are incapable of performing. What if he could harness that brain power and use it for the power of good?
Meanwhile, across the world, librarians are attempting to digitise books by scanning them into computers that are not always capable of recognising every word; either because the pages are old and yellowed, or because the ink is faded and blurred. So von Ahn combined these two problems to make a solution, and invented the ReCAPTCHA. Now, when you fill out a ReCAPTCHA on an online form, not only are you proving that you’re human – but you are also helping to put print books online. You, and Luis von Ahn, are the reason why thousands of classic books such as Sartre’s La Nausée are available for free on the Kindle.
ReCAPTCHA was acquired by Google for an undisclosed sum (it’s probably safe to guess it was in the millions), and von Ahn moved on to a new project using the same basic philosophy: Let’s use the Internet to gather together millions of human minds and use them to solve problems that computers can’t. That’s where Duolingo (finally) comes in. Most of the information available on the Internet is in English, and those who don’t speak English can’t access it, except often with Google Translate – and we all know how awkward and inaccurate machine translations can be. We also know why – because while you can load up a computer programme with information on the different possible translations of any one word, a computer doesn’t have the same notion of context that a human brain is able to process.
If I were to ask you, for example, what the word ‘set’ means, without any context, you wouldn’t even be able to tell me if it were a noun (a chess set) or a verb (‘Set the table’); never mind what kind of verb (a bowl of jelly does not set in the same way that the sun sets). Machine translation can take into account the linguistic environment surrounding a word, but it lacks the common sense that a human brain can use to determine the precise meaning of a sentence, and so Luis von Ahn set about creating a programme that would allow humans to fill in the blanks where translating the Internet was concerned, much as the ReCAPTCHA filled in the blanks left by computers when digitising books. With Duolingo, you begin by learning simple phrases in a language of your choosing (French, Spanish, German, Italian, Portuguese) and as you progress, you are given more complex sentences to translate – sentences that come from Internet sites that need to be translated.
But wait, you might say, how can I translate a language I’m still only learning? Well, if you’re given a sentence where you only know 9 out of 10 words, and you don’t know what the last one means, you can click on the word and receive a list of translations produced by a machine. Then you can look at the context and work out which one applies here. If your sentence says ‘He _____ the nail into the wall’ and you click on the missing word and are told it can either be a noun (‘hammer’, a tool used in DIY and construction) or a verb (‘to hammer’, the process of using a hammer in DIY and construction), you can tell which kind of hammer you need in this context: the verb. In this way, computers and humans work together to benefit the human (you get to learn a language for free) and the computer (its task of translating a website is aided), which of course entails other benefits for the world at large – the availability of masses of information in several languages, at no cost.
While translation in tiny segments all coming from different, unqualified translators may not breed the most idiomatic, cohesive translations of entire webpages, at least with this method there shouldn’t be too many actual errors (the quality control process is explained in the above video). More importantly, it’s an inspiring business model, and an ingenious outlook on the world. I’ve been using Duolingo to learn Italian for over nine months now, and the results for me have been fantastic. It’s no wonder Duolingo was chosen as iTunes’ App of the Year 2013 – the interface is both fun and simple to use, the results are fast and easily measurable, and what’s more, you never know what small corner of the Internet your translation might appear in.
P.S. No, this is not a sponsored post! It’s just a post about something I greatly like and admire.