The Four Criteria, and What Each One Rewards
Four criteria, four different questions being asked of the same 180 words. Knowing which one a sentence is earning marks under changes what you write.
The four criteria and what they weigh
Every CELPIP Writing response is scored on four criteria, each from 1 to 12, and the four are then combined by weight. Content and Coherence is 30 per cent. Readability is 25 per cent. Task Fulfilment is 25 per cent. Vocabulary is 20 per cent.
The weighted total is rounded to a whole number and reported as a level: 10 to 12 Highly Effective, 7 to 9 Effective, 5 to 6 Developing, 3 to 4 Limited, 1 to 2 Level M. The arithmetic is worth doing once by hand, because it makes something obvious that no amount of advice does. Moving Vocabulary from 7 to 9 lifts the total by 0.4. Moving Content and Coherence from 7 to 9 lifts it by 0.6, and moving Task Fulfilment the same distance lifts it by 0.5.
In other words, the criterion most candidates spend their preparation on is the one that moves the score least. That is not an argument for ignoring vocabulary. It is an argument for never sacrificing a bullet, a paragraph break or a correct tense in order to fit an impressive word in.
Content and Coherence, 30 per cent
This is the heaviest criterion and the least understood. It asks two things at once: are the ideas relevant, clear and supported, and do they arrive in an order a reader can follow without going back.
The content half rewards development. One reason explained with a consequence and an example is worth more than three reasons listed. The pool would be used by more people is a claim. The pool would be used by more people, because the arena is only open to registered teams while a pool serves families, seniors and school groups on the same afternoon is a developed reason, and the second version earns marks the first does not.
The coherence half rewards paragraphing and connection. One idea per paragraph, and each paragraph opening in a way that tells the reader where they now are. Coherence is not the same as connector words. A response stuffed with Moreover and Furthermore can still be incoherent if the paragraphs do not build, and a response with almost no connectors can be perfectly coherent if the ideas are ordered sensibly.
Readability, 25 per cent
Readability covers grammar, sentence variety, spelling, punctuation and paragraphing, and it is judged on one question above all others: do the errors get in the way of understanding.
That framing is the useful part. It means errors are not counted, they are weighed. A missing article in I sent email last week is visible but costs almost nothing, because the meaning is untouched. A tense error in When I will arrive, I call you costs more, because the reader has to stop and decide what you meant. Ten harmless slips damage Readability less than two that force a re-read.
Sentence variety sits here too, and it is the part candidates most often miss. Twelve sentences of identical length and shape reads as flat control even when every one is correct. Mixing a short declarative sentence against a longer one with a subordinate clause is what lifts this criterion, and it costs no extra words.
The safe rule when you are unsure is to write the simpler sentence correctly. A complex sentence with a broken clause scores worse than a plain one that works, because Readability rewards control, not ambition.
Task Fulfilment, 25 per cent
Task Fulfilment asks whether you did the specific thing you were asked to do. All instructions and bullets addressed. A clear purpose or a clear opinion. An appropriate tone. Proper email format on Task 1. One option clearly selected and supported on Task 2. Roughly 150 to 200 words.
It is the most mechanical of the four criteria and therefore the most avoidable to lose. Every failure here is a failure to read, not a failure of English. A bullet skimmed, a formal scenario answered informally, a Task 2 answer that presents both options evenly and never commits.
It also carries the fidelity rule, which is the single biggest cause of unexpectedly low scores in otherwise fluent answers. If the prompt says a glass cabinet was cracked and a box of kitchen items went missing, your email must be about those items. Writing about a broken television instead is a factual error, not a stylistic one, and fluent English does not excuse it. One substituted or invented fact, or one bullet left unaddressed, caps Task Fulfilment at 8 and Content and Coherence at 9. Two or more, or a wrong central premise, caps Task Fulfilment at 6.
Accurate paraphrase is not an error. Calling a box of kitchen items the kitchen box is fine. What is flagged is a genuine change of meaning: a different object, a different number, a different date, a different amount.
When you are unsure whether to write the complex sentence or the simple one, write the simple one. Readability rewards control, and a broken subordinate clause costs more than a plain sentence ever loses.
Vocabulary, 20 per cent
Vocabulary is scored on range, accuracy, collocation and fit with the topic and the tone. All four matter, and the last three are where marks are actually won.
Range means not repeating the same word four times in 180 words. It does not mean rare words. A response built from precise ordinary language scores well here, because precision is the thing being measured. Arrange for a plumber to inspect both units is better vocabulary than facilitate a comprehensive examination of the premises, and it is better for a reason the criterion states directly: advanced vocabulary used incorrectly does not score high.
Collocation is the quiet one. Make a decision and take a decision are both possible; do a decision is not, and errors of that shape are visible instantly. Fit with tone matters just as much. I would be grateful if you could is right for a landlord and wrong for a friend, where Could you is the better sentence.
How to use the weights while you are writing
Before you start, the weights tell you where to spend your planning: on what you will say and in what order, not on which words you will use. Content, coherence and fulfilment together are 80 per cent of the score, and all three are settled at the planning stage.
While you are writing, the weights tell you what to do when you get stuck for a word. Take the plain one and keep moving. Ninety seconds hunting for a synonym is ninety seconds not spent developing the bullet, which trades a 20 per cent criterion against a 30 per cent one.
At the end, the weights tell you what to check first. Bullets covered, register correct, option committed to. Those are Task Fulfilment, worth 25 per cent, and they are the fastest things on the page to verify. Spelling comes after, not before.
Check the prompt's own nouns against your answer before you submit. Substituting a single given fact caps Task Fulfilment at 8 no matter how fluent the rest is, and it is the most common cause of a lower band than expected.
Practise it
Five judgements about which criterion is in play and which version of a sentence earns more.