The Extra Mile
Robot owls: "The question that remains in my mind," writes Imogen West-Knights in the latest issue of The Dial, "and in the minds of many of even its most dedicated users, is: what is Duolingo actually teaching you?" (Emphasis mine.)
What indeed?
Duolingo won't make you fluent in a foreign language – even the company admits this – and frankly, I'm skeptical it will even get you to "conversational" levels of proficiency. And yet it remains the most popular language-learning app on the market, and millions of people dutifully do their Duolingo lessons every day.
Duolingo recently rolled out a new, more expensive tier to its subscription plan (the current one gives users endless "lives" so they needn't wait for more gameplay). Duolingo Super features an AI chatbot (of course), something that promises to address this flaw: the difficulty its users have in actually talking in the language they're spending so much time on. For $180 a year, Duolingo users will be able to "call" and "chat" with one of the app’s cartoons: Lily, the angsty teenage girl with the purple hair. Yes, the company purposefully chose her for her sass, all while repeating the claim you hear again and again from the AI-in-education evangelists: people feel more comfortable making mistakes with robots.
Will this new feature help? West-Knights isn’t so sure. Her conversations with Lily had “mixed results that only partly had to do with my stumbling French. The video call function is not impressive. Lily told me a story, asking if I was following, which I was. Then she asked if I had a funny story to tell, but before I had time to speak, she said goodbye and hung up.” But does that matter?
I think there’s something hopeful, maybe even touching, about all of us thinking that five minutes of language-learning a day is going to have us chatting away to natives. But that’s not the only thing you can get out of Duolingo. If you use the rowing machine in the gym twice a week for several years, does it matter that you never get very good at rowing on water? And whether or not Duolingo is a waste of your time, as many users I spoke to worried it might be, depends on what you go on to do with the limited knowledge it gives you.
Does it matter that Duolingo doesn't “work” — that it doesn’t really do a good job of teaching, that its users don’t really learn? I mean, of course it does – technologies both reflect and produce our understanding of the world around us. It matters when ed-tech sucks not simply because we aren't learning anything but rather because we are learning something. There are all sorts of lessons in even the shittiest piece of ed-tech about what teaching and learning "look like" and "feel like."
“What do you want to get out of this?”is a perfectly valid question — Duolingo asked me yesterday when I downloaded the app again, giving me a list of choices like “fun,” “travel,” “school,” and “talk to others.” sometimes our expectations, our imaginations have already been prescribed, and we don’t know that there are more possibilities than the multiple choice options offered.
Flaws in how language learning apps work (or don't work) isn't just something that’s a problem for Duolingo, of course. It functions much like its competitors – all digital flash cards, really – which, frankly, is how a lot of in-person, foreign language-learning classes operate too.
And there's a reason why these all tend to focus on phrases like “I’d like to buy a yellow hat” and "I'd like chicken with rice, please" and "where is the bathroom" – the history of foreign language instruction is wrapped up in twentieth century ideas about travel and tourism. And even that curriculum, that pedagogy has a longer history, one that can be traced back hundreds of years to Latin instruction (for the elite), with its focus on vocabulary, conjugation, declension, without any expectation for conversational fluency (for obvious reasons). So we should ask: what is any foreign language-learning teaching you — cosmopolitanism? Morality? Survival?
But Duolingo is also interesting, of course, because of its founder Luis van Ahn and his long history with artificial intelligence – well before Lily's eye-rolling appearance this fall. van Ahn is the inventor of reCAPTCHA, a CAPTCHA system now owned by Google, that was originally used as part of the company's massive book-digitization efforts. Wits reCAPTCHA, users were asked to verify they were humans and not bots, by deciphering hard-to-read text – a brilliant way to extract free labor for training the Google algorithms. van Ahn's original plans for Duolingo were something similar: the language-learning tasks that people would undertake on the app were to be used to translate the Web. (The company quickly ran into problems in the EU, where it is illegal to rely on unpaid translators; it dropped this crowdsourcing component, instead focusing on simply building an educational game.)
Why learn a foreign language anyway, particularly when there are tools like Google Translate — the ability to instantly translate (almost) anything with our phones. You can just take a photo of a menu, and it'll tell you which entree is "chicken and rice."
What happens to conversation if instead of learning to talk to others, we are content to just click alone? What happens to community when we need not struggle to communicate with one another and can instead just show someone our phone with a message ostensibly, perfectly transposed to their tongue?