How Reading Maps to CLB 1 to 12
A band is a description of how you read, not just a number out of 38. Knowing which behaviours sit at your target level tells you what to change.
A band is a behaviour, not a score
Your raw score out of 38 is converted to a CELPIP level from 1 to 12, aligned to the Canadian Language Benchmarks, and that number is what appears on your report and in an immigration application. It is easy to treat it as arithmetic and nothing else.
It is more useful to read it the other way round. Each band corresponds to a recognisable way of handling text, and your raw score is a measurement of which way you currently handle it. Two candidates at CLB 7 will have got different questions wrong and will still be doing the same three or four things.
That matters because it tells you what to practise. If your errors are the errors of the band below your target, the fix is a behaviour, not more hours. Almost nobody moves a band by reading more of the same material in the same way.
The lower bands: reading word by word
At the lower levels, reading is happening at the level of individual words and short phrases, and the effort of decoding consumes the attention that comprehension needs. A candidate here can often translate every sentence of a passage and still be unable to say what the passage argued, because nothing was left over for the argument.
The observable signs are consistent. Simple, short, concrete texts are handled reasonably. Anything with a subordinate clause, a pronoun referring back several sentences, or an idiom slows the read to a stop. Questions about explicit stated facts are answerable; questions about implication are effectively guesses.
Time is the binding constraint rather than accuracy. Parts are left unfinished, and the accuracy on the questions that were reached is much better than the total suggests.
The work at this level is not test technique. It is volume of ordinary reading at a level slightly below comfortable, daily, until decoding stops costing attention. Test strategy applied to a reading speed that cannot finish the part is strategy with nothing to work on.
CLB 5 and 6: finishing, but literally
Here the reading is fluent enough to finish most parts, and the errors change character. They stop being about not understanding and start being about understanding the wrong thing.
The characteristic behaviour is literal reading. What the text states is available; what it implies is not. Part 3, which asks only where a fact is, is often the strongest part at this level. Part 1, which hides its meaning in tone and omission, is often the weakest, which surprises candidates because the letter was the easiest thing to read.
The second characteristic is vulnerability to true-but-not-asked options. A candidate here recognises a detail from the passage, feels the recognition as correctness, and chooses it without rechecking the stem. Paraphrased correct answers get rejected for the same reason, because they do not feel familiar.
The move from here to CLB 7 is mostly one habit: reading the stem again, with the chosen option in it, before committing. It sounds trivial and it is worth several marks, because it is the only defence against the largest distractor family in the section.
Diagnose by error type, not by score. Unfinished parts mean the reading itself is too slow; attribution and copied-feature errors mean the comprehension is fine and the organisation is not.
CLB 7 and 8: understanding, but not tracking
At 24 to 30 correct, comprehension is genuinely good. Implication in Part 1 is mostly handled, paraphrase in Part 3 is mostly handled, and the passages are not the obstacle.
What separates this range from CLB 9 is tracking rather than understanding. A candidate here reads a Part 4 article, follows the argument, and reaches question 3 holding the positions without the people. They read a Part 2 chart, understand every bullet, and cannot say which column carried which feature. The information was comprehended and not organised.
The errors are therefore concentrated in specific places: attribution swaps in Part 4, copied-feature distractors in Part 2, and the second block of five questions in Parts 1 and 4, where the reply and the comment require holding a source text in mind while reading a new one.
Timing also stops being a background issue and becomes a limit. At this level there is usually enough reading speed to finish, and not enough spare capacity to also verify, so the last two questions of a part get answered without the checks that the first six received.
The behaviours that move this range upward are organisational and small: a ledger in Part 4, a discriminator grid in Part 2, naming the subject before evaluating options. None of them is a language skill.
Take three timed sections on different days before you believe a level. One passage on an awkward topic moves a raw score by several questions, so a single result is a point where you need a range.
CLB 9 and above: verification and discipline
At 31 correct and up, the remaining errors are rarely comprehension errors at all. They are the last few questions where two options both looked supportable and the more definite one was chosen, or where a second clause went unchecked, or where the clock ran out on a part that had been handled too carefully at the start.
The behaviours that characterise this range are unglamorous. Answers get verified against a named line. Options are read to the full stop. Strong words are checked against the text's own modality. Inferences are taken one step and no further, and when two options survive, the more cautious one is chosen.
Pacing at this level is deliberate rather than instinctive. The candidate knows the four per-part budgets, knows that the reply and the comment need more time than the stems that precede them, and abandons a resistant question on a schedule rather than on a feeling.
Above CLB 10 the difference is mostly consistency. The same disciplined reading applied to the hardest passage in the set, on a day that is going badly, without the quality dropping in the last part. That is a rehearsal question rather than a technique question, which is why full timed sections matter more at this level than at any other.
Locating yourself honestly
The raw ranges are worth knowing because they make the target concrete. Thirty-one or 32 correct is CLB 9, 28 to 30 is CLB 8, 24 to 27 is CLB 7, 20 to 23 is CLB 6, and 16 to 19 is CLB 5. Note how uneven they are. Three questions inside the CLB 7 range move nothing, while one question at the top of CLB 8 moves a whole band.
A single practice score does not locate you, because one passage going well or badly moves a raw score by several questions. Three sections, on different days, with the same timing, gives you a range rather than a point, and the range is the honest answer.
Then match the behaviours rather than the number. If your errors are attribution swaps and copied features, you are reading at the CLB 7 to 8 pattern regardless of what today's score said, and the organisational habits are what will move you. If your errors are unfinished parts, no amount of distractor theory will help until the reading itself is faster.
Finally, set the target above what you need. If you need CLB 9, that is 31 correct, so prepare to a standard of 33 or 34. The margin is not perfectionism, it is insurance against a passage on a topic you happen to find difficult.
Set your practice target two or three questions above the raw score your band needs. At the top of the scale the bands are two questions wide, and a margin is what absorbs one bad passage on the day.
Practise it
Five questions on reading the bands as behaviours, and on locating yourself among them.