CELPIP scoring rubric: what raters actually score

There is no single CELPIP scoring rubric. Two of the four sections are scored against one, and two are not. Writing and Speaking are judged by trained raters against four rated dimensions, each divided into five performance levels. Reading and Listening have no rubric at all: every answer is right or wrong, and a computer does the scoring.

That split is the most useful thing to understand about CELPIP scoring, because it changes what "getting better" means in each section. In Reading and Listening you are trying to get more answers right. In Writing and Speaking you are trying to give a rater more evidence to point at.

This page is the cross-skill view: what the two rubrics contain, where they agree, and how both routes end up on the same 0 to 12 scale.

#What is on the CELPIP Writing and Speaking rubric?

Four dimensions each, and three of the four are the same dimension in both skills. CELPIP publishes the categories and the factors inside them, so this is not guesswork.

The middle two rows are the only real difference, and the names tell you why. Readability is how easily your writing is read. Listenability is how easily your speech is followed. Same job, different channel.

Content / Coherence, Vocabulary and Task Fulfillment are identical in both, factor for factor, which has a practical consequence worth stating plainly: work you do on ideas, organisation and word choice counts twice. It is not Writing preparation or Speaking preparation. It is preparation for both.

#Where does grammar sit in the CELPIP rubric?

Inside the delivery dimension, in both skills, and never on its own. Grammar and sentence structure is a factor under Readability for Writing, and a factor under Listenability for Speaking. There is no grammar dimension anywhere on the CELPIP rubric.

That placement is a statement about why grammar matters. It is grouped with paragraphing and connectors in Writing, and with rhythm and pausing in Speaking. In both cases the question a rater is answering is how much work is it to follow this person, not how many errors can I find.

It also means grammar is one factor out of sixteen, sharing a dimension with three others. So "fix my grammar" is a smaller plan than it sounds. Our own data on Writing points the same way: across more than 70,600 anonymised, aggregated Writing responses from over 4,500 learners on HelloCelpip, about 35% of everything our evaluator flags is a Readability issue and about 27% is Vocabulary. Those are our evaluator's judgements on practice answers, not official CELPIP scores, but the shape is consistent: how cleanly you write and which words you choose account for most of what gets flagged.

The full breakdown, dimension by dimension, is in our guide to CELPIP Writing scoring.

#How does a rubric rating become a level from 0 to 12?

In two steps: raters assign a level in each dimension, then those dimension ratings are converted into your component level.

Each of the four dimensions is divided into five performance levels, and every level in every dimension has a written descriptor. A rater assigns you a level in a dimension by finding tangible evidence in your response that matches that descriptor. That phrase, tangible evidence, is the one to hold on to. Nobody awards a level for a general impression, which is why a safe, vague answer often scores lower than learners expect. There is less in it for a rater to point at.

Your component score is then derived from those dimension ratings and transformed into a CELPIP level. The conversion rules come from a standard-setting exercise, in which language experts and testing professionals decided what a test taker has to demonstrate to reach each level. That is the reason a Speaking 9 is meant to mean the same thing on your test form as on somebody else's.

Two facts about the people doing the rating, because they change how a borderline answer should feel:

So a response sitting between two levels is not rounded down by whoever opened it first. Disagreement triggers more scrutiny, not less.

#How are Reading and Listening scored without a rubric?

Dichotomously, by computer. Every Reading and Listening question is multiple choice or a similar format, and each answer is either correct or incorrect. A blank answer is scored as incorrect, so there is no reason to leave one.

Your raw count is not what gets reported. CELPIP uses many test forms, and although they are built to the same guidelines, they can still differ slightly in difficulty. Raw scores are corrected for those differences through score equating and then reported as a level, which is why the number of correct answers needed for a given level moves a little from one test to another. CELPIP publishes approximate conversions for both sections, and we cover them here:

• CELPIP Reading score chart
• CELPIP Listening score chart

The same equating logic is why no section reports a percentage, and why there is no total and no pass mark. Each of the four sections gets its own level from 0 to 12, and from level 3 upward that level is the same number as your Canadian Language Benchmark. If you want to see how four section levels line up against a specific programme, the score to CLB calculator does that, and CLB levels explained covers what each level means.

#What does the CELPIP rubric have no dimension for?

Quite a lot of what people worry about. None of these appears as a rated dimension or a published factor:

• Your accent as such. Pronunciation is a factor, but it sits inside Listenability beside rhythm and intonation, as part of how easily you can be followed. An accent a listener follows comfortably is not the thing costing you marks. There is more on this in our Speaking practice test guide.
• Your typing speed. CELPIP Writing is typed, and how fast you type is not a rated factor. How much you write is: word count sits under Task Fulfillment.
• Whether a rater agrees with your opinion. Task 2 and the opinion Speaking tasks are scored on how well you support a position, not which position you take.
• Who you are. Tests are assigned to raters randomly and your identity is never shown.

One thing that is on the list surprises people just as much: tone sits under Task Fulfillment, next to relevance and completeness. So writing a warm, chatty email to a city official is not a small stylistic slip. It is the same kind of problem as missing a required point.

#How do you use the rubric when you practise?

By naming the dimension every time you fix something. Most learners review an answer and think "that was bad". The rubric turns that into four separate, fixable things.

Here is one sentence from a complaint email, of the kind Writing Task 1 asks for, with four problems in it:

Four fixes, four different dimensions, one sentence. That is why a single-issue study plan stalls: you can clean up every grammar error you make and still have three dimensions nobody has looked at. CELPIP does not publish how the four are weighted against each other, so the safe assumption is that none of them is the one that does not matter.

The practical version of this is short. Write or record an answer, then read it back four times, once per dimension, and ask what a rater could point at each time. On HelloCelpip, Writing and Speaking questions come back with a report organised by those same four dimensions, with the specific lines quoted from your own answer, so you can see which one is actually costing you.

Try a Writing Task 1 email question →

Official sources:

• CELPIP Test Results
• CELPIP Score Comparison Chart
• Language requirements, Immigration, Refugees and Citizenship Canada

Last updated: August 10, 2026