The first year of a language is wonderful. Every week you can do something you could not do the week before: order a coffee, then read a sign, then survive a short conversation, then follow the plot of a film with subtitles. The progress is visible because it arrives as new abilities.
Then, somewhere around B1, it stops. You can handle most situations badly. You can be understood. And for months — sometimes years — nothing seems to improve. Most people conclude at this point that they have hit their ceiling, that they lack whatever other people have, and quietly stop.
They have not hit a ceiling. Five specific things happen at once around B1, and all of them are fixable once you can see them.
1. The words got rarer
Vocabulary obeys a brutal distribution. The most frequent thousand words of a language cover something like three quarters of everything anybody says. The second thousand adds maybe another eight per cent. By the time you are learning your six thousandth word you are buying fractions of a per cent of coverage per word.
So the arithmetic changes underneath you. Going from five hundred words to fifteen hundred transforms your life. Going from three thousand to six thousand — twice the work — produces a change you can barely feel day to day, because those words only show up in the specific situations where you previously changed the subject.
The learning has not slowed. The visible return on it has.
2. You switched from learning to compensating
This is the important one.
At A2 you cannot get around a missing word, so you have to acquire it. At B1 you have enough machinery to route around almost anything. You do not know the word for windscreen, so you say "the glass in front in the car". You cannot handle the subjunctive, so you rephrase until you do not need it. You avoid the past perfect for a year without ever consciously deciding to.
And it works. You are understood. The conversation succeeds.
That success is exactly the problem, because the feedback loop that drives acquisition is failure — the moment where you cannot say the thing. Once you are fluent enough to always route around the gap, you stop generating the failures that would have taught you. The gaps stabilise. Repeat that for long enough and errors stop being errors and become your permanent dialect; linguists call it fossilisation, and B1 is where it sets in.
Ask yourself which constructions you have not attempted in three months. That list is your plateau, written out.
3. Your ruler broke
Below B1 you measure progress by new capabilities, and they arrive weekly. At B1 you can already do everything, just badly — so there is no next milestone to cross, and "slightly less badly than in March" is not something you can feel.
You are probably still improving at roughly the same rate. You have just lost the instrument that used to show it to you. This is why the plateau is demoralising out of all proportion to what is actually happening.
4. Nothing is at your level any more
Comprehensible input — material a little above where you are — is the closest thing this field has to a consensus mechanism. It works, and at B1 the supply dries up from both ends.
Graded readers and learner podcasts are now too easy: you understand them completely, so they teach you nothing. Real native material — a novel, a podcast between two friends, a comedy — is still much too hard, dense with idiom and speed and cultural reference. The band that would actually stretch you has become narrow, and almost nobody publishes into it, because there is no market for material aimed at exactly your six-week window.
5. The syllabus ran out
Courses, apps and textbooks are overwhelmingly built for A1 to B1. That is where the customers are and where the material is easy to sequence: there is an obvious next tense, an obvious next thousand words.
Past B1 the curriculum fragments, because what you need next depends entirely on you — your domain, your errors, your goals. Most products simply stop, and you are left with no structure, no correction, and the word "immersion", which is advice in the same way that "be richer" is financial advice.
What actually gets you out
Ban your own workarounds. Take the list of things you avoid and make them compulsory. Retell a story and require the past perfect in it. Describe your week without the three verbs you lean on. It will feel like going backwards — slower, clumsier, more errors — and that is the correct sensation. You have deliberately removed the compensating strategy that was hiding the gap.
Go deep in one domain instead of broad across many. Trying to be B2 everywhere at once is a decade-long project. Being genuinely comfortable in one domain — your job, one hobby, the twenty conversations you actually have — is weeks, and it feels enormously better than a thin layer of everything. Then add the next domain.
Make output the centre, not the reward. Below B1 you can improve on input alone. Above it you mostly cannot, because the errors that limit you are in production, and reading more will never surface them. Speaking and writing is where the fossilised material becomes visible.
Get corrected, specifically. Not "that was great" and not everything at once. You need someone or something that notices the same error the fourth time and says so. Uncorrected output at B1 does not reliably make you better; it makes you faster at the version you already have.
Take native material and scaffold it, rather than waiting until it is easy. Same twenty-minute episode three times: once for the gist, once with the transcript, once for the ten phrases you want to steal. This manufactures the level of material that nobody publishes.
Change what you measure. Vocabulary counts are the wrong instrument now. Better ones: how many seconds of stall in a ninety-second recording, how many distinct tenses you actually produced, how many times you changed the subject because you could not say the thing. Those move while your CEFR level does not, and watching them move is what keeps people going through the flat part.
The unsentimental version
The rough consensus estimates for guided study hours look something like ninety to a hundred to reach A1, three to four hundred cumulative for B1, five to six hundred for B2, and perhaps eight hundred for C1 — for a language close to your own, and more for one that is not.
Read those numbers again. The step from B1 to B2 costs roughly as much as everything you did to get to B1 in the first place, and it delivers a change that is much harder to see.
So the plateau is not an illusion and it is not a defect in you. It is an accurate perception of a step that is genuinely two or three times longer than the previous one, walked without the milestones that used to mark the distance. The people who get through it are not more talented. They are the ones who stopped measuring the wrong thing, kept producing, and kept getting corrected while they did.