How (I think) I learn languages

An attempt to observe my cognition process from a distant position, then do some inductive reasoning. Just because there’s a lot of non-visual-first people out there.

tl;dr: When I’m curious about a new language, I pick up a few moderately long phrases, first test my ability to pronounce distinct syllables, then pronounce words, then, gradually, the whole phrase—all to myself, in my head—until I believe they sound good, and then I find a victim to test my new skill on. Only afterwards do I proceed with vocabulary and grammar, if ever necessary.

Now, deconstruct.

First, phonetics

If you’ve ever tried to transcribe someone’s speech into text, you could have noticed how bad their grammar actually was. Or maybe not actually bad, but sometimes just inconsistent. That’s because brain processing time dedicated to making up a phrase between a thought and an action is too short to produce perfect structured flow on the fly. There are just so many processes running in speaker’s head, many of them in the background. You can be sure they didn’t even notice when they made this or that mistake.

That’s because what matters most is phonetics. How you sound is more important than all the complex syntactic structures you use and well-chosen idioms you masterfully use.

At the same time, phonetic load of any message you hear is what makes you, for example, guess where that person is from or detect a lie quickly. Phonetic patterns are the cause, accents and emphases are the effects, and our brain is good at scanning it and analyzing both, mostly unconsciously.

So I always try to learn what kind of phonetic development native speakers have. Does the language they speak require little amount of mouth movement, or does it involve many different parts? Does the way the mouth opens and shuts matter, are there many “soft” sounds, is there any difference between the same sound pronounced in the beginning and in the end of a word?

It’s mostly trying to mimic certain phonetic patterns native speakers naturally follow. And to facilitate it, there’s a (childish) thought experiment: what if I was Chinese and so I found it really hard to pronounce “r” sound?

Loan words don’t count though. They spoil the learning. Common Western “ninja” and Japanese 忍者 (also “ninja”) are pronounced differently. But it shouldn’t affect the mimicking process at all, and when learning Japanese, I have to incline towards the latter in 100% cases.

So, generally, what really matters is sounds and syllables in action. Without understanding any meaning behind them at all. It’s actually quite entertaining. After the phonetic system of a foreign language has been learned to a certain point, the next step is learning words.

Second, memorize phrases and deduct into words

The next thing after picking up phonetics of learning a language “for real”. Meaning, trying to remember and integrate into speech some phrases and words. Mostly phrases.

On phonetic level, phrases are indistinguishable from words: they are just chunks of voice that begins, ends, and produces some sounds in the middle. On words and phrases level, phrases are actually easier to understand than words, and therefore they should be more preferable in the learning process. Take two phrases, mix them, and see if they make sense. Multiply by number of phrases you know while continuously adding new phrases to your list and—BAM!—you speak the language!

If you take closer look at how kids learn to talk, right after they break this silence barrier, then you’ll see they just repeat syllables. They don’t know that syllables alone don’t make sense to adults, so they struggle, sweat, and gradually develop to a point when they can pronounce something that sounds as a whole word.

I’m pretty sure that, inside a toddler’s head, there’s a complete mess but also there’s a hunger for immediate feedback: I told you this, will you respond positively or not? Toddlers are kind of in the early epoch of training their speech model. Through constant feedback, they train it quite fast—until they can say a short phrase.

Once kids have learned short phrases, and not just words, it’s so clear that they simply repeat what they’ve heard most. Which always proves to be the most reliable tactic.

As kids grow up, they start learning their language, reading books, learning new patterns and testing them against broader circle of people: kids from the neighborhood, classmates, teachers, parents’ friends. At this point, the system is so huge, parents can hardly predict anything: Where did you learn that word? Which is great, but also sometimes a disaster, when your kid comes home and asks what is “fuck” for the first time.

Then, as adults, their network of communication contracts. Ours has been doing it already, too. We create families, get hired into companies or create businesses, spend time at the same coffee shops, or bars, or libraries, with the same people. Our expansion slows down, until it eventually halts. At that point, we learn new things at a much slower portion over what we reinforce.

There’s totally that friend of yours who’s always saying that phrase, right? Reinforcement.

So, what I do is I try to break from the comfortable pattern, listen carefully, compare what I already know, and then apply. I believe it’s very annoying, and people think I’m weird or maybe I’m up to something. But geez, I can’t help it. Sorry, guys!

Finally, reverse-engineer down to grammar

Only when I know enough phrases, I use deduction process to understand grammar. It blends well with traditional learning process: books, tutors, MOOCs, whatever.

I used to start with books from day 1. It never worked. Except one thing that worked pretty well: kids’ books.

Definitely worth using them to learn new languages, there’s nothing wrong in it. In any kids’ book, the story is usually short, the plot is simple and linear, and the phrases are very typical. For some time, you’ll sound like a kid who’s always speaking using quotes from books, but that’s okay, compared to suffering from learning the language “adult way”.

One little trick I always use is the one that makes me possibly the most annoying person in the world: I learn a phrase, learn another phrase, compare them, and then, of course, I’m not getting it. So I look for a native speaker around, and if there’s none (there’s never anyone), I go outside and ask any stranger my question. They are amused, but they never say “no”. Who’d ever say “no” when the question is about their own native language, essential part of their culture? (hint: works well in southern part of Europe)


Just as in anything else, there are phases of divergence and phases of convergence. During the former, I aimlessly grope for new information; during the latter, I focus on learning next one thing. I believe it works the same way for pretty much anyone else. Still, it’s worth putting it directly: whenever progress gets stuck, repeat the same steps again.

It takes great effort to become an expert, but it takes just a little effort to know a language well enough. Academics call it A1. And it’s actually never about going from A1 to B1 or anywhere.

The real problem is going from zero to A1. So this little framework, “phonetics → phrases → understanding”, I hope, might happen to be useful to somebody out there.