How Speaking Maps to CLB 1 to 12
Your speaking level is the rounded average of four criterion scores, and it maps onto the Canadian Language Benchmarks almost one for one. What changes between the bands is more concrete than most descriptions suggest.
How the number is produced
Each of the four criteria, Content and Coherence, Vocabulary, Listenability and Task Fulfilment, is scored from 1 to 12. They carry equal weight, and your speaking level is the rounded average of the four. There is no weighting towards fluency and no bonus for any single criterion.
The levels are grouped into named bands. Levels 10 to 12 are described as Highly Effective, 7 to 9 as Effective, 5 to 6 as Developing, 3 to 4 as Limited, and 1 to 2 as Level M. Most candidates sitting the test in earnest land somewhere between 6 and 9, and the bands above that describe near-native control rather than a slightly better version of a good answer.
Because the number is an average, it is unusually sensitive to your worst criterion. Ten, ten, ten and six averages to nine. Nine across the board also averages to nine, from a considerably less lopsided starting point. If you are close to a band boundary, the arithmetic says to find the lagging criterion rather than to improve the strong ones, and the caps described elsewhere in this course are the usual reason one criterion is lagging.
The CLB mapping, including the part charts skip
From CELPIP level 3 upwards the mapping to the Canadian Language Benchmarks is one to one. CELPIP 9 is CLB 9, CELPIP 10 is CLB 10, and so on to 12. This is why immigration requirements quoted in CLB translate directly into a CELPIP speaking level with no conversion table needed.
The bottom of the scale is where it stops being clean, and most published charts quietly omit it. CELPIP level 2 corresponds to CLB 1 to 2, a range rather than a point. CELPIP levels 1 and 0 are recorded as not assessed, meaning there was insufficient information to place the performance rather than that the performance was poor.
For the levels that matter to most candidates, the descriptions are worth reading literally. CLB 9 is effective proficiency in some high stakes contexts. CLB 10 is highly effective proficiency in some high stakes contexts. CLB 7 is adequate proficiency in somewhat demanding contexts, and CLB 8 is good proficiency in more demanding contexts. The progression is about the difficulty of the situation you can handle, not the number of mistakes you make.
What separates CLB 7 from CLB 9: vocabulary
The vocabulary ladder is the most concrete of the four, and it explains a lot of plateaus. Around band 5 the entitlement is common words and phrases only, with no expectation of abstract language. Around band 9, context-specific vocabulary appears, along with some idiom. At 11 and 12 there is a broad range of both concrete and abstract language, and figures of speech arrive naturally.
The step from 7 to 9 is therefore not about difficulty of word. It is about specificity and abstraction. A candidate describing a pharmacy who uses place, people, things and working is speaking generally. One who uses the dispensing counter, a repeat prescription, the queue for the till and restocking the shelves is speaking specifically, and no word in that list is advanced.
The abstraction half is the same move seen elsewhere in this course. People were annoyed is concrete. What annoyed people was the sense that nobody was in charge of the queue puts a layer above the physical scene. A candidate who can do that reliably in every task is describing what they observed rather than listing it.
What separates CLB 7 from CLB 9: structure and flow
The Listenability ladder is control first, then range. Around band 5 there is some control of simple structures. Around 7, good control of simple structures. Around 9, some control of complex ones. Around 11, good control of a broad and diverse range of complex structures.
This is the ladder that surprises careful speakers. Producing flawless simple sentences is exactly what a band 7 looks like, so a candidate whose strategy is to keep every sentence short and safe has designed themselves a ceiling. The step to 9 requires attempting subordination, relative clauses, conditionals and embedded questions, and landing most of them. Some control means most, not all.
The flow half of the criterion is measured directly from the audio, and it is the part you can move fastest. A comfortable rate is roughly 120 to 160 words a minute. Filler density under two per hundred words is unobtrusive, and over six per hundred caps Listenability at 6. Any single silence over four seconds caps it at 7. Those are not descriptors, they are thresholds, and a candidate sitting at 7 because of a five second pause is not a 7 speaker.
Find your lowest criterion before you decide what to practise. The level is a rounded average of four equal scores, so one criterion at 6 costs you a whole band while three at 10 cannot buy it back.
What separates CLB 7 from CLB 9: audience
The Task Fulfilment ladder runs convey, then adjust, then adapt. Around band 5 you can convey information on a familiar topic. Around 9 you adjust your style and tone for a range of different audiences. Around 11 you adapt to the situation, the purpose, and your relationship with the listener, all three together.
This is why the tasks keep naming who you are speaking to. A cousin, a supervisor, a father, a co-worker, a neighbour. Those names are not colour, they are the material this criterion is scored on, and an answer that would sound identical whoever it was addressed to has demonstrated conveying rather than adjusting.
Adjusting looks like small, consistent choices. Contractions and look and honestly with a friend. You might want to consider and full forms with somebody senior. Softening an ask to a stranger and being more direct with a sibling. Adapting, the next step up, adds the purpose and the relationship at once: acknowledging that you have asked this person for something before, or that the request is awkward precisely because of who they are to you.
Aiming at the right number
Work out the level you actually need before you decide how to prepare, because the routes to different targets are different. If you need a 7, the fastest route is almost always removing caps: finish every task, use the time, make the request, take a position, address the right person. None of that requires better English.
If you need a 9, caps alone will not get you there. You need the vocabulary step, which is specificity plus abstraction, and the structural step, which is attempting complex sentences and landing them. Both of those are habits built over weeks rather than adjustments made on the day.
Above 9 the descriptions change in kind rather than degree. A 10 is an answer with no weakness you could name, and 11 and 12 describe control that is effortless to follow across every task, with figures of speech and calibrated certainty appearing naturally. That is a level of English, not a test technique, and preparing for it as though it were a technique is the most common way to waste a month.
One practical note on targeting: because the four criteria are averaged and then rounded, being consistent is worth more than being brilliant. A candidate who is excellent on three tasks and abandons one is averaging a strong performance with a weak one across every criterion at once.
If you are aiming at 7, work on removing caps rather than on your English: finish every task, use most of the allowed time, make the request, take a position. If you are aiming at 9, the work is specificity, abstraction and complex sentences you can actually land.
Practise it
Five questions on how the levels are built and what actually changes between them.