This might seem like a strange topic for a blog post by an economist, but it relates to the “correlation is not causation” theme that is ever-present in economics. I think adults are actually much better at learning languages than kids, despite contrary correlations (disclaimer: I’m not talking about pronunciation/accents here, which may well be a different story). Allow me to elaborate.

I can’t count how many times I’ve heard people claim that kids are, all else equal, better at picking up new languages than adults (this blog post is a recreation of a discussion I’ve had many times). Observationally, this is certainly true – people who move to new countries as kids generally speak the language well, while those who come as adults often have smaller vocabulary and incorrect grammar. Case in point is my family. I moved from Ukraine to the US when I was 12, and my sister was 10. It took us about 6 months to move from ESL (English as a Second Language) to a regular English class (we had learned some English in Ukraine as well), and our written and spoken English is essentially indistinguishable from a native speaker, although some people claim I have a slight accent. By contrast, my parents to this date make grammar mistakes and sometimes struggle with vocabulary. Their English is way better than it used to be, but far from flawless. This pattern is very typical of an immigrant family, in my experience.

It’s very easy to conclude from this strong age-at-arrival correlation that younger kids are just inherently better at learning languages. But economists like to benchmark most situations against the “ideal experiment”. If your goal is to determine whether younger individuals are better at learning languages, the ideal experiment would be to take some adults and some kids, place them in exactly the same new-language situation for a period of time and then estimate the relationship between new language knowledge and age at the end. You can alternatively use observational data to deduce the likely results of such an experiment, but only if you can control for any other factor that affects language acquisition and is also systematically different between people of different ages, other than age itself. If you can’t control for some relevant factor, you have an omitted variable problem, which can reduce the quality of your results.

Another look at my family’s experience suggests some potentially important omitted variables. First, my sister and I literally had no other job but to learn English (ok, maybe we also washed the dishes and walked the dog). For at least 8 hours a day, we were in an environment where everyone around us spoke English. I would then come home and watch TV with closed captions on (very useful for learning a new language) and do homework in English. And then there’s the desire to fit in. Not being able to express myself to same-age peers was quite uncomfortable, and I was very motivated to remedy that quickly. When someone made fun of my pronunciation of “th” in 7th grade (I pronounced it like “z”), I learned how to pronounce the sound correctly within a few days.

My parents’ environment was very different. My dad was in a company that employed many Russian speakers (in case you’re wondering, my family is from Kiev and are all native Russian speakers) . He had a full time job (where, similarly to many immigrants, it was not his English skills that he was being paid for) and all the adult responsibilities you can imagine to worry about. My mother, who eventually went to work for the same company, also had many responsibilities that did not involve engaging with English speakers. And, like many immigrants, my parents quickly found other Russian speakers to spend their free time with. And as far as I can tell, there was no external pressure for them to become fluent.

Given these systematic differences between younger and older people, all of which would bias us toward concluding that kids are better language learners, it’s difficult to trust any study that claims to tease out the effect of age on language acquisition from observational data on fluency and the age at which a person started learning a language. But I think there are other observational data that make it pretty clear that adults are better language learners than kids. It takes a baby about a year to say its first word and another year to form simple sentences. My native-born almost-6-year-old’s sentences are riddled with grammar mistakes and her vocabulary is quite limited. Frankly, her English is still pretty bad. Imagine that you immersed a 30-something year old into a completely foreign family in a foreign country that treated him or her the same way they would treat a baby (language-wise: speaking slowly, using simple words, enunciating) and that 30-something had no responsibilities at all. It doesn’t matter how different the language is, I doubt that it would take him or her more than a week to pick up some basic vocabulary or more than a month to start using simple sentences. I took French for several hours a week in high school, and by the end of the four years my French was better than my almost-6-year-old’s English. And I was still spending the vast majority of my days speaking and learning in English (yes, technically I was still a kid, but I bet I could do it again if I had that kind of time). While this is also observational evidence, I can’t think of any omitted variables that would overturn this line of reasoning (feel free to disagree in comments).

Finally, it turns out that my claim that adults are better language learners than kids is not that novel. Exhibit A: link1, link2, link3, and a Gizmodo article. But I’m writing this post anyway because I think this is a point that people are often wrong about. So if you want to learn a new language as an adult, don’t let the fear that you’re not as good at learning as a younger person stop you. What should stop you is the fact that you have a ton of responsibilities and no one to support you while you take an hour or two per day to learn a new language 🙂