The bands describe behaviour, not knowledge. Every descriptor in the framework is a statement about something a person can do with a language, and not one of them names a tense, a structure or a word. So there is no B2 grammar to study, and every list of grammar by level you have seen was compiled by a publisher rather than by the framework.
That is not a technicality. It changes what the scale is for: it is an instrument for describing where somebody is, and it was never a syllabus, so asking it what to learn next gets you a description of the destination and no route.
Eight facts about the framework
- the framework covers all languages, not just English, which is why the same labels appear across every branch of this pillar.
- can do statements define each level. A level is a description of behaviour rather than a body of grammar.
- A1 to C2 in three pairs: A is basic, B is independent, C is proficient.
- A2 handles routine exchanges, shopping, directions and simple personal information.
- B1 copes with familiar situations and can describe experiences and give short reasons.
- B2 follows an argument, understands technical discussion in their own field, and interacts with fluency and spontaneity.
- C1 uses language flexibly for social, academic and professional purposes, and this is the level most institutions mean by fluent.
- C2 is not native speaker. It describes precision and ease across every register, and a native speaker with no education in the language may not reach it.
Written for languages in general, which is why it names nothing
The framework was built by the Council of Europe to describe ability in any language, so it could not mention the perfect tense or the subjunctive without becoming a document about one language. What it could do was describe performance: whether somebody can follow an argument between two speakers, or write a letter of complaint, or sustain a conversation past the point where they run out of prepared material.
The consequence for a learner is that the same B1 statement is satisfied by completely different knowledge depending on the language. A B1 speaker of English and a B1 speaker of Japanese have almost nothing in common in terms of what they have memorised, and they can do the same things.
Six bands hiding a finer scale
The six letters everyone quotes are a compression. The framework itself describes plus levels, so A2 plus and B1 plus are real positions in it, and they matter because the plus levels are where most learners spend most of their time: the distance from B1 to B1 plus is a year for many people, and reporting it as B1 for both years is why progress feels invisible.
The 2018 companion volume extended it further, adding descriptors for mediation, which is relaying between two people or two texts, and for online interaction, neither of which existed in the original scale and both of which are a large share of what people actually use a second language for.
Where it stops: searching for the grammar list
The stall is looking for something the framework does not contain. A learner decides on a target band, searches for what to study to reach it, and finds a list of structures labelled with that band. The list looks authoritative, it is on a coursebook site, and it was assembled by an editor who had to put something in the syllabus.
Working through a publisher's list is not useless, and it is aimed at nobody in particular. The descriptors that a particular learner fails are personal: two people at the same band fail different statements, so no fixed list can be pointed at either of them. The list-following learner ends up studying structures they already had while the six things actually holding them at the band go untouched.
The correction is to work from the descriptors and not from the list. Read the band above yours, mark the statements you could not perform today, and treat those as the syllabus. It will be six or eight items and they will be nothing like the published one.
Where the level of a word or a structure was actually measured
The English Vocabulary Profile is the honest version of the list people are looking for, free after a registration, and its distinction is that it was built from evidence rather than from editorial judgement: it reports the level at which learners were observed using a word or a particular meaning of it, drawn from a very large collection of learner writing.
It replaced the publisher's level list, which is one editor's opinion presented with the framework's authority attached to it. The limitation is exactly the one the framework has: it tells you where a word sits in the observed order of acquisition, and it cannot tell you which words you personally are missing, which is still a question only your own performance answers.
What it looks like when it goes wrong
The recognisable failure is a learner who can name their band and cannot name a single thing they cannot do. The band was a label rather than a measurement, usually inherited from a class they were placed in, and it survives for years without ever being tested.
The second is quoting one number for four skills. A single band is an average across reading, listening, writing and speaking, and the four are rarely within one band of each other, so a person reporting a single letter is reporting the highest of the four in almost every case.
The unequal size of the steps between these bands, and the hours behind each, is in where you are now, and what the next step actually contains, and finding out which band you are honestly in, rather than which you were labelled with, is placing yourself honestly without paying for an exam.
What the framework will not settle for you is which band anybody needs. That question is answered by an institution with a number on a page rather than by a scale, and the number is often lower than learners assume: what each test scores, and how to prepare for yours.