How to Learn Any Language: The Full Fluency Formula Method

There is one specific moment that sends people searching for how to learn any language, and it is almost never a test.

It is a dinner table. Everyone is talking, fast, and none of it is aimed at you. Then someone turns, her father maybe, and asks you one question. And everything you were sure you had, the 300 day streak, unit 14, the year of classes, is not there. You freeze. Someone translates for you. The conversation moves on without you.

I know that moment because I sat in it for a long time.

Nine years of French immersion as a kid. Three years on Japanese with a strict immersion approach. Then I met the woman who is now my wife, whose family speaks Mandarin, and none of it helped me at their table. My day job is chief operating officer of a mid eight figure e-commerce company, which means I get paid to find systems and make them more efficient, so I pointed that at Mandarin instead. I did it in secret. 182 days later I blurted out Mandarin mid conversation at that same table, and her father froze and pointed at me. Later I asked her parents for her hand in marriage, in Mandarin. The longer version is on my about page.

I just put the whole method on YouTube as a free course. Six lessons, one hour and forty six minutes, the whole Fluency Formula method, six lessons. This post is the written companion, same order, so you can skim it or come back to one piece.

The language was always the vehicle. What you want is a seat at a table, and everything below exists to get you there if you want to get fluent enough to sit down and belong.

You were never bad at languages, you were on the wrong scoreboard

Here is the reframe that everything else hangs on, and it is lesson one.

You were not losing all those years. You were winning. You kept the streak, you hit unit 14, you got the grade, maybe even the level. By every number anyone put in front of you, you were doing great.

You were just winning a game that was never pointed at the table.

Look at what you were told to count. Streaks, which are days you opened an app. Units and levels, which are badges. Grades, which are proof for a file. Fifteen minutes a day, which is a dose. I have had students with 2,000 days logged in one app who still were not close to where they wanted to be. Five and a half years of showing up, aimed at the wrong target, which is the case I made about why your language streak is lying to you.

Now ask who decided what counts.

An app wins when you open it tomorrow. A class wins when you pass on Friday. A government agency wins when you hold the certificate. Not one of those dials has a setting for you at the table, and it would be a strange business model to build something a customer graduates out of.

So they worked. They measured activity, not ability.

Here is the split underneath it. Studying builds one thing in your head, and speaking needs a different one. Rules, tables, word lists, conjugations, all of that builds knowing about a language. You could read a physics textbook containing the actual laws that describe balancing a bike on two wheels. It will not get you down the street.

Running the language is the other side: following it live, answering without translating in your head, catching the joke in time. None of that gets a badge.

So what number predicts a conversation? Hours of the language you mostly understand.

Not hours studied. Hours understood. And notice how neatly that number stayed invisible, because nobody can sell you a subscription to it. No certificate, no level. Just the feeling of being able to talk to the person you care about.

Nothing you have ever owned has reported that number to you. Not once, by anyone. Which is exactly why it was never your fault.

Everyone quits a language in the same week, and it isn't willpower

Work with thousands of learners across a lot of languages and you start seeing patterns. This is the strangest one, and it is lesson two.

Different people, different languages, different years. Same week.

Week one feels like flying. New method, new energy, everything moving. Then around week four or five it stops. You are still doing the reps, but you cannot feel the needle. By month two it is quietly gone. You never decided to quit. One day became two, two became a week, and you never picked it back up.

Then the sentence arrives. Maybe I'm just not a language person.

Notice the timing. It never shows up in week one. It only shows up after an attempt has already fallen over. It is not a diagnosis, it is an alibi, which is why I think there are no language people at all, only methods that held and methods that did not.

It is not talent, and it is not a life stage. People ask me about children and neuroplasticity constantly. Here is what a child actually gets: somewhere between 10,000 and 20,000 hours of input handed to them for free. My son Pierce is currently figuring out that he has toes. He is not running an efficient protocol, he is just living, at enormous volume, for years.

You have three things he does not.

Concepts. You already know what the object on my head is, so when you hear the French chapeau or the Mandarin word for it, you are pairing a sound with a concept you own. A child is building both at once.

Patterns. If I say "yesterday I bought a pizza at the," you already know the next word is not "skateboard."

And choice. A child sprays attention at everything. You can aim.

So there is no talent column. There is an hours column.

What actually moves the quit point is felt progress backed by a number. A show you could not follow last month that you can follow today. A joke you laughed at before you had time to translate it. Those do more for your follow through than any pep talk, because they are evidence.

I have a student named Chris who came in calling it his last shot before giving up for good. Nothing about Chris changed. The spot moved. Another one, Steve, was learning Portuguese. Just past the 100 hour mark he posted that he had watched something and understood 95 percent of it, starting from zero. Three months later he was following native Brazilians at full speed. He did not get better at languages. He got hours that counted. Your own number will move with your language, your background and the hours you actually put in.

The catch is that the spot does not move on its own. It needs the right hours, at the right level, every week. Two missed weeks and you are back on the old line, and the curve flattens exactly where it always flattened, which is most of what people mean by the plateau problem. That is not weakness. That is gravity.

And the people who got past it did not have more willpower than you. Their hours were chosen for them, and somebody was watching whether the hours happened. If you are learning your partner's language, that second part is sitting right next to you. Tell them.

The only hours that count when you learn a language

Lesson three is the one I would keep if I had to throw out the rest.

Start with what a language actually is, because this is where most methods go wrong at the root. You do not recall a language. You run it. Name the last time you conjugated a verb in your own language. You did not. It just ran.

So the question stops being "what should I memorize" and becomes "what installs the system."

The answer is hours you mostly understand. The technical name is comprehensible input, which just means content you can mostly follow because the pictures, the context and the situation hand you the meaning of the words you do not know yet. The term comes from the linguist Stephen Krashen. I will say it once and then use plain language, because the term matters far less than the dose.

And there is a dose. This is the part nobody gives you a number for.

Under about 30 percent understood, it is noise. Over about 90 percent, you are just watching television, which is lovely but not buying you much. The zone is roughly 30 to 50 percent, call it 30 to 60 at the edges. I used to put that window around 75, and the more people I work with the further down I move it, though I would not call the higher end wrong. That one idea, that you only need to understand about half and your brain does the rest, is the whole engine.

Now hold everything you have already tried against that window.

The textbook and the class. Knowing about a language, graded, on a curriculum that does not care what you find interesting.

The daily streak. Activity inside an app, not time engaged with the language.

Subtitles in your own language. This one stings, because it feels productive. Your brain takes the easiest path to meaning, always, and if that path is sitting right there in English it will read the story and never run the language. Also, gently: there are no subtitles at your in-laws' dinner table. More on that in why you're using subtitles wrong.

The podcast in the car, half heard. Passive immersion has a place at higher levels, but early on it does little. And no, you are not learning a language in your sleep.

Videos about the language, which is exactly what you are reading right now. Somebody said that given eight hours to cut down a tree, he would spend seven sharpening the axe. Fair. But the axe is not the tree.

Flashcards, done as pure recognition and never run in context. Most people conclude flashcards do not work for them. Done right they are one of the highest returns on your time, which is the argument behind the flashcard hack most polyglots miss.

Every one of those landed outside the window. So take this from it: you cannot really fail a language, you can only quit it. Your hours were mis-aimed, not wasted.

Two more things about the window, because they are what makes this hard in practice.

It slides. Per language, per level, per person. What sat perfectly in your window on day one is at 90 percent two months later and stops paying you. A beginner playlist lasts roughly a month. The window is the easy part. Finding content that fills it, month after month, interesting enough that you do not resent it, is the actual job. Nobody grinds through a documentary on Irish horticulture just because the comprehension level is right.

And front load frequency. The top 1,000 to 1,500 words make up roughly 75 to 80 percent of what people actually say. Put those in your head early, on purpose, then let comprehensible input place them into context so the system can run them. That is also why skipping the polite beginner ladder works better than it should, which I unpacked in the toothpaste paradox.

I learned this the expensive way. In Japanese I did a lot of raw input with none of the accelerators, and a lot of it was me banging my head against a wall at a speed I did not have to accept.

You can't speak a language better than you understand it

This is the sentence I hear more than any other, and it is lesson four: I understand but I can't speak.

You follow the show. You catch most of what is said around you. Then you open your mouth and nothing comes out the way you meant it. So you book the tutor, you do the speaking challenge, and come out more convinced that you are bad at this.

You were not failing. You were early.

Speaking reads out a model. It does not build one. Input builds the model of the language in your head, quietly, on autopilot, and speech is that model read aloud. You cannot pour out of a glass that was never filled.

Firing early costs you something real. Say it wrong a thousand times before the model exists and the thousand and first time is wrong too. Mistakes become habits, and habits are hard to undo.

So there is a floor under every sentence you will ever say. You speak at the level you understand, usually a bit below it, never above it. Comprehension is the ceiling. Speech is the floor. The floor follows. It never passes.

Which means the freeze is not a courage problem. When you lock up mid sentence, part of what you are reaching for sits in the gap above your ceiling. You know what you mean in English, the model does not have that piece yet, and the mouth stalls. A comment on my channel put it better than I could: the words don't come, and when they do, the grammar is all wrong. That was me at my in-laws' table in Mandarin. Even on the night it worked, I spent the evening twisting around the things I could not yet say.

Same reason a tutor or an AI cannot rescue you below the floor. They will drill with you forever, and what they are drilling is the wrong version of you. That money was not wasted. It was spent early.

Then the floor rises, and it moves fast. One of my students, Gus, banked a few hundred understood hours and went to Mexico for two weeks. Day one he could barely ask for water. Day three, four and a half hours at a table in Spanish with his girlfriend and her family. His mouth just had to catch up, and that took days, not months.

And you cannot feel your own ceiling from inside it. The week before you are ready and the week after feel identical. It is a number, not a feeling, which is why somebody else usually has to call it for you and say: you have the hours, start talking.

Both sides of that line cost you. Too early and you freeze, and every rep trains the wrong version. Too late and you stall, and the family stops waiting. Inside the window, speaking badly is not a failure. Speaking badly is the plan.

Fluency is a date, not a dream

You will love me or hate me for this one, and lesson five does not soften it. Fluency is arithmetic.

Somebody told me once: work is killing me, I have only had time to do flashcards. Two years of slow and steady, and the table still felt a decade away. He was not lazy. He was never given a number.

So here is one.

Every language has a price in hours, because every language sits at a different distance from English. Italian runs around 600 hours. Mandarin or Korean closer to 2,200. Those are classroom hours counted by a government, aimed at carrying yourself professionally. Understood hours run faster, and conversational fluency arrives earlier than professional proficiency. Treat those figures as a ceiling, not a sentence.

Why can you not just stack 15 minutes a day until you hit the number?

The bucket. I am not a neuroscientist, but the way I understand it, your brain consolidates during sleep, moving what seems important into long term memory. Fifteen minutes never clears that bar. It leaks out faster than you pour it in, so you are not building, you are forgetting at a polite pace. That is the whole reason 15 minutes a day won't make you fluent, even when you never miss a day.

Which brings me to the advice I would like to personally retire. Slow and steady wins the race. In languages, slow and steady is the quit. It is months of grinding where nothing clicks, then eight years on and off, restarting from near zero because the bucket drained in between. You stop before anything paid you back.

So here is the division. Take the hours your language costs and divide by the hours a day you can actually give it.

600 hours at three hours a day is 200 days. Under seven months, which is roughly the 200 day path.

600 hours at one hour a day is 600 days. Under two years.

600 hours at 15 minutes a day is 2,400 days. Six and a half years, and that assumes nothing leaks, which it absolutely does.

Same language, same person, three completely different lives. Nobody ever sat down and did that arithmetic with you.

So front load it. Sprint while it is still work, and get to where native content takes over, because after that it is not studying, it is watching the next episode of the murder mystery at the end of a long day. The steep part is short. The flat part is free.

Andre is the version of this that still impresses me. Italian, 61 days, and the whole motivation was to propose in Italian in front of her family. He did it, posting under a fake handle from Assassin's Creed the whole time so his girlfriend would not spot him in the community. Mine was learning Mandarin for my wife's family: 182 days, no tutor, no classroom, no trip to China. Neither of us waited for years to pass.

And here is what the circled date costs. A date means a sprint. It is fun for a few days and then it is grueling, and you wake up earlier than you want to. Every previous attempt died at week five, where felt progress stopped. The people who got through did not have more grit. Their hours were pre chosen, so they never had to decide what to do that day, and somebody was watching the calendar with them. Nobody finishes a sprint unwatched.

You're fluent when you pass the table test

Ask 99 people what they want fluency for and 99 of them describe a table. Lesson six names it precisely, because you cannot aim at a finish line you cannot define.

Nobody ever cried at a certificate. They cry at a dinner.

So here is the test, and it has no score.

Can you say what you mean? Not a rehearsed sentence. Your actual thought, including the joke, including the awkward story about walking in on Uncle Tommy in the bathroom.

Can you follow what is said? Not just your partner slowing down for you. The whole table, at their speed. People adjusting for you all night is exhausting for them, and you can feel it.

Can you get through a day without bracing for it? No quiet dread before the family lunch. You just go.

Three questions, no level attached to any of them, because nobody is 43 percent at a table. You are on the inside or you are outside. That is why I never got excited about the A2 to C1 ladder. Is cracking a joke back at your mother in law a B1 thing or a C1 thing? The scale cannot see what you came for.

My first pass was Mandarin in front of her parents. People I had been around plenty and never actually known. That evening paid for every hour.

And it is not just me, which is the part that matters. Four tables, four languages. Andre proposed in Italian on day 61. Gus did hours at his girlfriend's family's table in Spanish with no English. Chris, on what he called his last shot, used Mandarin in Beijing. Oscar and Katerina spent two hours with a farmer who spoke no English at all, entirely in German.

Not one of them sat a test. Jobs, kids, commutes, no spare time, same as you. Not one polyglot on that wall. They were never language people, they just had a plan, and you can hear real students tell it in their own words.

How to learn any language, in the order that actually works

So what do you do this week? The whole thing, compressed.

Name the table. Not a level, not a certificate. The specific room, the specific people, the conversation you are shut out of. That is the only target still motivating you in week five.

Price it and circle a date. Look up the hours for your language, divide by the hours a day you can honestly give it, and put the date where you will see it.

Front load the first 1,000 to 1,500 words by frequency. Flashcards are fine here, as long as you are retrieving rather than recognizing, and those words land back into real content quickly.

Get your hours inside the window. Content you understand roughly 30 to 50 percent of. No subtitles in your own language. Interesting enough not to resent. Replace it about once a month.

Do not speak yet. Bank hundreds of understood hours and let the model build. When the ceiling is high enough, the floor comes up behind it faster than you expect.

Tell somebody. Ideally the person whose language this is. The people who get past week five are not tougher, they just were not alone.

The one thing six lessons cannot hand you is the day by day order: what goes in your window this month at your level, who calls it when you are cleared to speak, and who watches the calendar so the date holds. That last piece is the fluency roadmap, and it is where Fluency Formula does the work a video cannot. Nobody finishes this alone. Not even me. My wife found out halfway through my Chinese experiment, and I told her the rest myself.

Questions I get asked about how to learn any language

How long does it take to become conversationally fluent in a language?

It depends on the language and on how many hours a day you give it, and both are numbers you can look up and choose. Italian runs around 600 hours and Mandarin or Korean closer to 2,200, and those are classroom hours aimed at professional proficiency, so conversational fluency shows up earlier. At three hours a day, 600 hours is 200 days. At 15 minutes a day it is over six years.

Why can I understand a language but not speak it?

Because speaking reads out a model that comprehension already built. When you freeze mid sentence, some piece of what you are reaching for sits above your ceiling, so your mouth has nothing to read from. The fix is not more speaking practice. It is more input you mostly follow, which raises the ceiling and drags the speaking floor up behind it.

What is comprehensible input?

Content in your target language you can mostly follow, where the pictures, the context and the situation hand you the meaning of the words you do not know yet. The term comes from the linguist Stephen Krashen, and it is the only kind of hour that reliably installs the system. Aim for roughly 30 to 50 percent understood. Under 30 is noise, over 90 is just television.

Do I need to learn grammar to speak a language?

Not in the way you were taught. You already follow rules in your own language you could never name: you say the big red dog and never the red big dog, and you cannot tell me what that rule is called. You absorbed it from thousands of hours of understood language, not from a table, which is why I argue you should learn grammar through input. Grammar can clarify a mistake you keep repeating. It does not build the system that runs in a live conversation.

Is Duolingo enough to become fluent?

An app is genuinely excellent at one thing, which is taking you from no interest in a language to some interest. But it counts activity inside the app, not hours of language you understood, and those are two different numbers. I have had students with 2,000 days logged in one app who still were not close to where they wanted to be. Use it for the spark, then move your hours into input you mostly understand.

How many hours a day do I need to study a language?

More than 15 minutes. At that dose most of it is gone by the next morning, so you add slower than you forget. If your life allows two or three hours a day for one season, take it and race to the point where native content carries you, because after that the maintenance is watching what you would have watched anyway.

How do I learn my partner's language?

Start by naming the table, because that is the real goal, not a level. Price the language in hours, divide by what you can give it each day, and circle the date. Fill those hours with content you understand about a third to half of, and hold off on speaking until you have banked hundreds of understood hours. Then tell your partner, because the language barrier with in-laws closes faster when somebody at that table knows you are coming.

Am I too late to learn a language?

No, and the advantage is not where you think. A child gets 10,000 to 20,000 hours of free input, but they are building the concepts and the words at once, while you already own the concepts and can aim every hour at what works. If you are following this sentence in English, you have the kind of brain that can do this. There is no talent column. There is an hours column.

The last thing

If you have tried this before and it died, nothing here is an accusation. You did the reps. You showed up, often for years, which is the hardest part of this and the part most people never build. You were just pointed at a scoreboard that could not see the table.

So point the same effort at understood hours, put a real date on it, hold off on speaking until the model is there, then go say something badly to people who will be delighted you tried. That is the whole method, and it works the same whether you want to talk to your in-laws in their language, propose in Italian, or get fluent enough in Japanese to sit in an izakaya and let an old guy tell you his life story.

The full course is free and sitting right here: The Fluency Formula Method: How to Learn Any Language. Watch it in one sitting if you can.

And if you want the day by day order, and someone watching the calendar with you, Fluency Formula is where the path gets walked with help.