Duolingo as memento mori

About a month ago, I took a weeklong work trip out to California. It was great: a perfect mix of productivity and pleasure. One of my strongest memories of the trip was of walking to the office, listening to Prince, and feeling intensely grateful.

When I got back to DC, I was greeted by an angry owl. Duo, the strigiform mascot of Duolingo, emailed me to say that I hadn’t done my Spanish lesson in five days and that my 317-day streak was now but a memory. Over the next week or so, Duo seemed to pass through several stages of grief. He was sad. He tried bargaining (“Do you need a break?”). But my routine had been reset and I couldn’t get myself to go back to my lessons.

These emails had presented me with an opportunity to think about why I’d started learning Spanish in the first place. It was never a very practical idea. Partly, I was motivated to study Spanish so I could finally complete a long-standing goal of attaining fluency in another language, which itself stemmed from insecurity around being functionally monolingual. More generously, I’m always fascinated by the process of watching myself gain some new faculty: how there’s always some new previously invisible distinction — some neglected layer of information in the world — that I can suddenly detect.

Having gone through the process of gaining toddler-level fluency in other languages — French, German, Mandarin, and (get this…) Esperanto — Spanish seemed like a reasonable choice to see myself go farther than I had before. Spanish would at least have a certain ubiquity to it that I could lean on to keep me engaged, I thought. It still seems like the most sensible choice of any language to me for exactly that reason, but I’m not convinced anymore that learning a language is a sensible use of my time.

There are two mental models through which I’m looking at this question of whether to return to learning Spanish. The first is sort of a developmental suitability lens; the second a question of whether our time is best spent strengthening our innate talents or shaping ourselves into well-rounded individuals. Both of these could probably be subsumed under the umbrella of expected utility, but I think it’s worth teasing them apart a little before we deal with them together.

First, we behave differently at different ages. Part of this is due to a negotiation with our cultures about what is appropriate as one grows older, but part is due to biological processes that change our capacities with age. Our brains tend to lateralize over time, meaning that we tend to become more extreme versions of whoever we were before. Critical periods in skill attainment close at some point, too, after which reopening them is difficult or sometimes seemingly impossible.

Part of why we change our thinking is also that the costs and benefits of learning new things change over time. For each day that we wait to learn a new skill, the learning takes up a larger share of our remaining life and the benefits have less time over which to accrue. That ratio just gets more and more skewed over time, and it’d be irrational to forget about it. This is the driver behind Alison Gopnik’s claims that the brain faces the same explore-exploit trade-off as machine learning algorithms do. When we’re young, we try out all sorts of things just to find out where we might have the best success of discovering reward. As we age, though, it becomes harder to justify so much exploration, and we tend to settle into a pattern of exploiting the known avenues of success that we learned long ago.

I have some problems with this framework when the world or our utility function shifts, but with a thing like the Spanish language, which I could have learned more easily and profitably with a younger brain, I think the central insight stands: i.e., not to be dramatic, but time is running out for me (us!), so keep an eye on the ever more hyperbolic ratio of costs to benefits.

The other issue is this question of whether it’s better to focus on skills where you have a natural edge, or to try to ensure that you don’t have any major weaknesses. The latter is tempting. It’s embarrassing to not know things that you think you should know — like a second language — and so we tend to assume that it’s worthwhile to spend time fixing those deficiencies. Maybe for some people and some deficiencies it is, but the success of the whole Gladwellian 10,000 hours perspective was based on selling people the myth that we can achieve mastery of virtually anything with enough effort, when instead the distinctions between those with innate talent and those without just become more apparent in the upper eschelons of performance.

Phrased another way, this question is trying to get us to understand what life is asking of us. I’ve been encountering this idea a lot, that life asks us to respond to it by reshaping ourselves — or maybe I’ve just been noticing it more. It’s a nice corrective to the perhaps more romantic idea that our inner essence is some eternal thing that would be the same across time or space. The question of what life asks of us shifts us instead to a view of self-expression as a conversation with our times and our surroundings.

Which brings me to the question: does the world really need another data scientist who can speak mediocre Spanish? I don’t think so. Nor do I think that learning it gave me enough pleasure that it was worth pursuing for that reason alone. If it had, I wouldn’t be asking this question. I’ve always wanted to get something out of it.

At the same time, it doesn’t feel like it’s enough to just say goodbye to my dream of bilingualism. I feel like I have to make a trade and recommit myself to something that my Spanish learning had made me skimp on. Writing is a top contender. The more I do what I call genre writing, like for grants or journal articles, the more I feel the need to balance that with the voice I use here. To stretch my lexical legs, as it were. That’s a deal I could feel good about making with myself and you: I get to be monolingual forever, you get to read more of my thoughts. Sound good?