The Duolingo app has revolutionised the way people learn a language, and now the company has set its sights on language certification. CEO and co-founder Luis von Ahn tells The PIE News how he hopes the 20-minute exam taken on a mobile device can offer a cheaper and more secure alternative to industry standards.
The PIE: What year did you launch?
LvA: 2012. We now have 120 million users and we have a bunch more really cool statistics. For example, there are more people in the US learning languages on Duolingo than there are people learning languages in the whole US public school system.
The PIE: How many languages do you offer on your app?
LvA: 21. It’s a huge range from English and Spanish to tiny languages, we teach Catalan and Irish.
The PIE: A lot of people will know Duolingo, but what’s your story?
LvA: This is my third company. I’m a professor of computer science at Carnegie Mellon University. I sold my two previous companies to Google. The reason I started this company was so that it would be mission driven for education.
The PIE: Do you have a background in education? What was your inspiration for the education mission?
LvA: Well, I’m a professor but also I’m from Guatemala and when I was starting Duolingo I wanted to do a project on education but my views on education are very related to where I’m from. It’s a very poor country and a lot of people talk about education as something that brings equality to the social classes, but I always saw it as the opposite – something that brings inequality to people.
“I wanted to do a project on education but my views on education are very related to where I’m from”
What happens, particularly in poor countries, is the people who have a lot of money can buy themselves the best education in the world. They can come here to the US to get degrees and because they’re so well educated they end up continuing to have a lot of money. Whereas the people who don’t have very much money barely learn how to read and write, and therefore never make a lot of money. I wanted to do something that would provide equal access to education to everybody and that’s where Duolingo came up.
The idea is it’s a free way to learn languages.
The PIE: How big is the company now?
LvA: About 85 people. Mostly based in Pittsburgh.
The PIE: Why did you decide to develop an English language test?
LvA: A couple of years after we launched we starting getting a lot of emails that all said the same thing: ‘Thank you for teaching me English, I wasn’t able to afford it before, but now I have a problem, I need to certify that I know English.’
So we looked into the English certification business and we found that it was pretty crazy. There’s about five billon dollars a year spent on it. The way people certify their language skills is through standardised tests. And they’re all pretty similar – they all cost about $200, you have to go to a testing centre to take it somewhere, and the whole process takes four to six weeks.
Most of the people who have to certify that they know English are in developing countries and there, 200 bucks is a month’s salary
That sounds annoying but it’s even worse because most of the people who have to certify that they know English are in developing countries. And there, 200 bucks is a month’s salary, plus the testing centres are not in every city so you have to travel. The whole thing is this crazy ordeal which seems like something that was made in the 1960s, which it was.
We thought we could do a lot better, so we launched an app where you can basically take a standardised test for English. Now we’re also trying to certify it and our goal is to become a global certificate for it. Will we be able to? Who knows, but that’s our goal.
The PIE: What’s been the reaction to your exam so far?
LvA: We have piloted with some universities in the US, a few this last year – Tufts, Santa Clara, Yale, MIT – and we’re going to get more this year. The way we see it, it’s really good for the applicants, for the students themselves, so if the universities ever start offering us in lieu of any of the other tests, most everybody will just want to take our test because it’s cheaper and you can take it from anywhere you want.
The PIE: How does it solve the problem of security and cheating?
LvA: The reason you have to go to a testing centre is to make sure that you are you and you aren’t cheating. The way we do security is we turn on the front facing camera on your device or microphone and a real human proctor watches the whole testing session. They get access to what you’re looking at on your screen, the whole screen, and also a live video of you. It turns out to be really hard to cheat with this. What cheating looks like is this [moves eyes up and down] because they’re reading somewhere. It’s really easy to tell.
When you’re on boarding to start the test we also have you show us your ID. But probably, the biggest thing about security for us, is at the end of the test we have a five minute free-form interview. We say, ‘hey, you already took your test, now just answer these three questions.’ They’re open-ended questions. We do a few of these and then your test certificate includes a link to the video of you speaking. And that’s really hard to cheat because you have to speak and in the end if it doesn’t look like you or the guy who shows up on campus doesn’t look like you, that’s pretty obvious.
“We’re way more interested in correlating with actual English speaking ability”
The other thing that’s pretty good is when you look at how much of the cheating happens in standardised tests, a lot of times it’s the testing centres that have been compromised because usually these are in developing countries – essentially people get paid off to let them cheat. Here you can’t quite do that because you can’t see the proctor.
The PIE: How valid are the exam scores?
LvA: So as a first level, what we would like to do is correlate with the TOEFL and the IELTS. And it turns out we have pretty high correlation with those, because there have been a couple of studies that were done, one by the University of Pittsburgh, one by somebody at DePauw. But then we’re way more interested in correlating with actual English speaking ability. So for example, the DePauw study found it wasn’t correlating with the standardised test, it was correlating with what ESL teachers at the end of the first semester gave the people as a grade. And it turns out that we had higher correlation than the TOEFL and the IELTS for that.
The PIE: How can a test given on a mobile device test the four core skills of language acquisition? Do students actually write essays on their mobile?
LvA: They don’t write an essay. They do speaking, the have to do listening, they have to do reading. For the writing component, we tested with a bunch of vocabulary and we find that’s pretty highly predictive of their writing skills. One of the reasons we don’t do writing is its very hard to grade. They essentially hire people who get paid $6 an hour to grade it. It’s not very good, you can’t do it with a machine but you’ve got to grade I don’t know how many millions of these. That’s the ugly truth of how writing tests get graded. So that’s why we don’t do it.
“One of the reasons we don’t do writing is its very hard to grade”
The PIE: How are you planning to eventually compete with IELTS and TOEFL worldwide?
LvA: It’s the newer technology and it’s the thing that belongs in this century. Not having to go to a testing centre. Not having to pay so much money. Our hope is that that will be enough. And prestigious universities are starting to use this. Harvard is already using it for some of their departments. The idea is if it’s good enough for these universities, it’s good enough for others. It won’t be quick but hopefully in a few years this will become the standard.
The PIE: How much is your test?
LvA: $50. It’s a quarter of the price. And part of the reason we have to charge is we have to pay somebody to actually watch the person taking the test.
The PIE: What’s next for Duolingo?
LvA: It’s the most widely used language learning app, but we’re going to try to get in to other types of education. The next thing we’ll probably do is learning how to read and write. There are a billion adults in the world who don’t know how to read and write. Fifty million of them have smart phones, so we would like to teach them all how to read and write. That would be a decrease by 5% in global illiteracy. If we’re able to do that, it would be pretty major.