CELPIP Speaking in 2026: All 8 Tasks, and the Real Bar

Most people preparing for CELPIP spend their time on Reading and Writing, because those feel like the parts you can study. The exam body's own numbers suggest that is the wrong allocation.

Paragon publishes a mean Speaking score of 7.76 on CELPIP-General. Add up the band percentages in the same report and only about a third of candidates reach band 9 or above. Speaking is where the distribution thins out.

If you want to build the habit that actually moves that number, SpeakShark is free with three AI conversation sessions a day and no card, each capped at five minutes and four turns.

In this guide: what CELPIP Speaking is · no human in the room · the eight tasks · the four with no IELTS twin · what the scores show · how it is rated · how to practise · General or LS · what we could not verify · how we researched this guide · FAQ

Key takeaways

CELPIP Speaking is 15 minutes and 8 tasks, one question each, per the official test format page.
There is no human interviewer. CELPIP is fully computer delivered and the official page states there is no separate speaking session.
Paragon's 2024 Data Report gives a mean Speaking score of 7.76 on CELPIP-General and 6.67 on CELPIP-General LS.
Adding the published band percentages in that report ourselves gives roughly a third of CELPIP-General candidates at Speaking band 9 or above, and about 78% of LS candidates below band 9. Both totals are our arithmetic, not Paragon claims.
On CELPIP-General LS, speaking is one of only two scored bands, so it carries half your result while taking about a fifth of the roughly seventy minute sitting.
Comparing the two published task lists ourselves, four of the eight are not set as discrete tasks by IELTS, so IELTS-only preparation leaves half the test unrehearsed. Neither exam body publishes that mapping.

What CELPIP Speaking actually is

CELPIP is the Canadian English Language Proficiency Index Program, administered by Paragon Testing Enterprises. It comes in two versions and both contain the same Speaking component.

CELPIP-General covers all four skills and takes under two hours fifty minutes in total: Listening 46 to 55 minutes, Reading 43 to 56 minutes, Writing 53 minutes, and Speaking 15 minutes, per the official test format page. Paragon's 2024 Data Report gives wider windows for Reading and Writing, so treat the format page figures as the planning numbers and expect some variation on the day.

CELPIP-General LS covers Listening and Speaking only and runs about seventy minutes: Listening 46 to 55 minutes plus the same Speaking 15 minutes. Speaking is therefore about a fifth of your time in the chair, but it is one of only two scored bands, so it carries half your result. That imbalance is unusual, and it is why this version deserves separate thought rather than being treated as a shorter General.

Scores are reported in bands from 0 to 12, mapped onto the Canadian Language Benchmarks.

Which version you need depends on what you are applying for. The official Who Accepts CELPIP page lists the accepting bodies, and IRCC's own help centre names both CELPIP General and CELPIP General-LS among the tests it accepts for citizenship, adding that it does not accept any other third party test results even if they are similar. Check your own programme against those pages rather than against any guide, including this one.

Nobody is in the room with you

This is the single structural fact that reshapes preparation, and it cuts both ways.

CELPIP is fully computer delivered. You read a prompt on screen and then speak into a microphone. The official test format page states plainly that there is no separate speaking session. No examiner sits opposite you, nobody nods, and nobody looks confused when a sentence goes wrong.

If interviews make you freeze, this helps enormously. A large part of IELTS Speaking anxiety is social, and CELPIP removes the social part entirely.

If you feed off a listener, this hurts. Many fluent speakers rely on the small feedback loop of another person's face to know they are being understood. Talking into silence for fifteen minutes is a skill of its own, and it is not one you build by chatting with friends.

That second point is worth taking seriously, because it is trainable and almost nobody trains it. Our guide to practising English speaking alone covers the specific problem of producing sustained speech with no listener, and overcoming the fear of speaking English covers the version of this that shows up as a blank mind.

The eight tasks in order

The official test format page names all eight. One question each.

#	Task	What it asks you to do
1	Giving Advice	Advise someone on a situation
2	Talking about a Personal Experience	Narrate something that happened to you
3	Describing a Scene	Describe a static image
4	Making Predictions	Say what happens next in that scene
5	Comparing and Persuading	Weigh two options and argue for one
6	Dealing with a Difficult Situation	Handle an awkward social problem
7	Expressing Opinions	State and defend a view
8	Describing an Unusual Situation	Describe something out of the ordinary

Read that list as a set of functions rather than topics. The topics change every sitting; the functions do not. Advising, narrating, describing, predicting, comparing, persuading, negotiating a social problem, and opinion giving are eight distinct things your English has to do, and you can rehearse all eight without ever seeing a real prompt.

We have deliberately not reproduced any real test content here. Question content belongs to the exam body, and Paragon publishes official free sample material for anyone who wants to see the real thing.

The four tasks IELTS does not set

If you already prepared for IELTS, half of CELPIP Speaking is familiar and half is not. That asymmetry is the most useful thing in this guide.

Describing a Scene (Task 3) asks you to describe a static image in detail. IELTS has no equivalent. The nearest analogue anywhere is PTE's Describe Image, and even that is usually a chart, graph, map or table rather than a social scene you have to read for context. We cover this one in full in CELPIP Task 3, Describing a Scene, including why it must be prepared together with Task 4, and our PTE Describe Image guide covers the general skill of talking about a picture under time pressure.

Making Predictions (Task 4) follows straight on from the same scene and asks what happens next. Prediction as a language function is not unique to CELPIP, and IELTS Part 3 can push you toward it. What is CELPIP specific is predicting the next state of one particular image you have just finished describing. This is hypothetical language under time pressure, and it is a specific grammatical gear most learners rarely use out loud. Because it works from the image you just described, it fails when your description left nothing unresolved, which is why we treat Tasks 3 and 4 as one problem.

Comparing and Persuading (Task 5) asks you to hold two options in mind, weigh them and argue for one. This is the loose overlap rather than a clean gap. IELTS Part 3 touches comparison, but it rarely asks you to commit to a recommendation and defend it as a task in its own right.

Dealing with a Difficult Situation (Task 6) is the one people underestimate. You are handed an awkward social problem and asked to talk your way through it. This is closer to a workplace conversation than an exam answer, and it rewards politeness strategies, softening and repair, none of which appear on a vocabulary list. Role play scenarios and English job interview role plays are the closest practice format we have for it.

The practical consequence: preparation built only on IELTS material leaves four of eight tasks unrehearsed. If you switched tests to avoid the interviewer, that is a fair reason, but do not assume the preparation carries over.

What the published scores actually show

Paragon publishes a 2024 CELPIP Data Report, and it is more candid than most exam bodies' statistics.

Mean Speaking score: 7.76 on CELPIP-General, 6.67 on CELPIP-General LS. The LS mean sitting lower is worth noticing, since LS candidates are typically applying for citizenship rather than permanent residence.

The band distribution on CELPIP-General is the number that should shape your study plan. The report publishes the percentage at each band: 13.1% at band 9, 7.1% at 10, 7.9% at 11 and 4.5% at 12.

Add those and you get roughly 32.6% at band 9 or above. That total is our arithmetic on Paragon's published table, not a figure Paragon states, so treat it as a calculation rather than a quotation. What it means is plain enough: about two thirds of candidates finish below band 9 in Speaking. Doing the same addition on the LS table gives about 22% at band 9 or above, so roughly 78% of LS candidates land below band 9, again our arithmetic on Paragon's table rather than a figure Paragon states.

Two more figures from the same report give useful context. 96.0% of test takers report their purpose as applying for Canadian immigration or citizenship, so this is not a study abroad test with an immigration side door; it is an immigration test. And 80.5% of candidates are aged 20 to 40.

The report also states the test is delivered at over 140 test centres in Canada and internationally.

How your recording is rated

Your responses are recorded and assessed after the fact. Paragon's 2024 Data Report states they are evaluated by at least three trained and certified raters against standardised scoring rubrics.

Three things follow from that.

There is no examiner to charm. Nothing you do with eye contact, warmth or apology affects the score, because the person scoring you was not there.

Audio quality is part of your performance. A rater can only score what they can hear. Speaking too quietly or too close to the microphone is a self inflicted wound in a way it never is with a live examiner.

Consistency matters more than peaks. With multiple raters and standardised rubrics, one brilliant answer among eight does less for you than eight solid ones. This is an argument for breadth of practice across all eight functions rather than perfecting a favourite.

How to practise for eight unrelated tasks

Drill the functions, not the topics. Take any subject and run it through all eight: advise, narrate, describe, predict, compare and persuade, handle a complication, opine, describe something unusual. The topic is disposable; the function is what is scored.
Practise talking into silence. Record yourself speaking with nobody reacting, and get comfortable with it. This is the specific condition CELPIP creates and it is unlike ordinary conversation.
Rehearse the picture tasks with real pictures. Open any photo, describe it out loud until you run dry, keep going past the point where it feels finished, then predict what happens next in it. Tasks 3 and 4 are the pair most often skipped in preparation.
Work on repair language for Task 6. Softening, apologising, proposing an alternative, checking the other person is happy. Office small talk in English and how to talk on the phone in English cover the same register.
Fix clarity before range. Raters score what they can understand. Six pronunciation fixes in fourteen days targets the sounds that block comprehension, and reducing mother tongue influence covers the patterns underneath them.
Speak daily rather than long. Fifteen to twenty minutes of unscripted talking beats a weekend of theory, because what is measured is retrieval under pressure. Free methods to raise fluency, building confidence in English and a 30 day plan to improve speaking cover the habit.

Not sure where you are starting? Test your English speaking level free first, and if the honest answer is that you have never spoken English under time pressure, read why your English speaking is not improving before booking anything. When you are ready to put the reps in, start a free session and run one topic through all eight functions.

General or General LS

Structurally the choice is simple and the consequences are not.

CELPIP-General is all four skills. CELPIP-General LS is Listening and Speaking only, with the identical eight speaking tasks, in about seventy minutes.

Because speaking is one of only two scored bands on LS, a weak speaking day has nowhere to hide. It is still only fifteen minutes of that roughly seventy minute sitting, about a fifth of your time, but it carries half your result. On CELPIP-General a strong Reading or Writing result sits alongside your Speaking band; on LS it does not exist. The lower published mean on LS, 6.67 against 7.76, is consistent with that.

Which one you are eligible for and which one your application requires is not something a blog should tell you. IRCC's help centre names both, and the official Who Accepts CELPIP page lists the routes. Read those.

If you have not settled on CELPIP at all yet, CELPIP versus IELTS for Canada PR covers that decision, and it turns on one structural difference rather than on which test is easier.

What we could not verify

We would rather leave gaps than fill them with numbers we cannot stand behind, so here is what is missing and why.

We are not publishing a candidate volume figure. No official annual CELPIP test taker count appears in any Paragon or Prometric publication we could reach. The 2024 Data Report publishes percentages only.

We are not publishing the CLB thresholds for specific immigration routes here. The Express Entry minimums, the post graduation work permit thresholds, the citizenship requirement and the validity period of a result all live on canada.ca, and they change. Numbers for those circulate widely and we are not going to repeat them second hand in a test format guide. We have since read those pages directly and set out the Express Entry minimums, the two moment validity rule and the citizenship conditions in the English test for Canada immigration. An earlier version of this paragraph said canada.ca could not be loaded from our environment. That was wrong: it was a limitation of one fetching tool, not a fact about the site.

We are not quoting a test fee. The official site publishes Canadian prices, states that international pricing varies, and its own IRCC specific page carries different figures from its test format page, so the site is internally inconsistent on this. Check the price for your own country at booking.

We are not listing which professional regulators accept CELPIP. The exam body publishes a list, but we have not verified individual regulator pages, and some professions are notably absent from it. If you need CELPIP for a licence, confirm with that regulator directly.

How we researched this guide

The test structure, the component durations, the eight named speaking tasks and the absence of a separate speaking session come from the official CELPIP test format page. The mean scores, the band distribution, the 96.0% immigration purpose figure, the rater process and the test centre count come from Paragon Testing Enterprises' 2024 CELPIP Data Report, read directly. The list of accepting bodies comes from the official Who Accepts CELPIP page, and the citizenship test list comes from IRCC's own help centre.

The roughly 32.6% figure for band 9 and above on CELPIP-General, and the matching figure of about 78% below band 9 on CELPIP-General LS, are our own addition of the band percentages Paragon publishes, not numbers Paragon states, and we have labelled them as such wherever they appear, including in the meta description.

We have not reproduced any test content. Paragon publishes official free sample material, which is the right place to see real prompts.

Practise speaking, from SpeakShark

SpeakShark is an AI English speaking practice app, and it is our pick for the kind of speaking practice CELPIP demands, for one specific reason: the test asks you to speak into silence, without a listener reacting, across eight unrelated functions. That is a strange condition, and the only way to get comfortable with it is to produce a lot of unscripted speech. You talk, the AI answers what you actually said, and you get speaking feedback while the conversation is still going. The free tier gives basic feedback; the detailed pronunciation and grammar breakdown is on Premium.

Start free with three sessions a day and no card, each free session capped at five minutes and four turns. Paid sessions run ten minutes with unlimited turns. Limits are on the pricing page, and how it works walks through a full session.

We are a speaking improvement tool. We are not an exam preparation provider and we are not affiliated with Paragon Testing Enterprises, Prometric, IRCC or any exam board. For test format, rules, booking, official sample material and the requirements of your own immigration route, go to the exam body and to canada.ca directly. Use SpeakShark to make your spoken English stronger, and use official material to learn the test.

Site: speakshark.com
Support: speaksharksupport@gmail.com
Free tier: 3 AI conversation sessions per day, no card

Sources

CELPIP test format: component durations, the eight speaking tasks, and the absence of a separate speaking session
Paragon 2024 CELPIP Data Report: mean speaking scores, band distribution, the 96.0% immigration purpose figure, the rater process, test centre count
Who Accepts CELPIP: the official list of accepting bodies and routes
IRCC help centre, accepted language tests: CELPIP General and CELPIP General-LS named for citizenship
Australian Department of Home Affairs, Competent English: CELPIP General approved for Competent English

FAQ

How long is the CELPIP Speaking test?

Fifteen minutes, per the official CELPIP test format page, containing eight tasks with one question each. Paragon's 2024 Data Report gives a slightly wider window of 15 to 20 minutes. It is the shortest component of CELPIP-General, which runs under two hours fifty minutes in total, but on CELPIP-General LS it is one of only two scored components, so your speaking band carries half your result even though it is only fifteen minutes of a roughly seventy minute sitting.

Is there a human interviewer in CELPIP Speaking?

No, and this is the biggest structural difference from IELTS. CELPIP is fully computer delivered: you read a prompt on screen and reply into a microphone, and the official test format page states there is no separate speaking session. Your recordings are then evaluated afterwards. Paragon's 2024 Data Report says responses are assessed by at least three trained and certified raters against standardised rubrics.

What are the 8 CELPIP Speaking tasks?

In the order the official test format page lists them: Giving Advice, Talking about a Personal Experience, Describing a Scene, Making Predictions, Comparing and Persuading, Dealing with a Difficult Situation, Expressing Opinions, and Describing an Unusual Situation. Comparing the two published task lists ourselves, four of those are not set as discrete tasks by IELTS, and Part 3 touches the ground under only one of them. Neither exam body publishes that mapping, so treat it as our comparison rather than an official one. It is why IELTS preparation transfers less cleanly to CELPIP than most people assume.

What is a good CELPIP Speaking score?

It depends entirely on what you are applying for, so check your own programme rather than chasing a number. What the data shows is where people actually land. Paragon's 2024 Data Report gives a mean Speaking score of 7.76 on CELPIP-General and 6.67 on CELPIP-General LS. Adding up its published band percentages ourselves gives only about a third of CELPIP-General candidates at band 9 or above, which is our arithmetic on Paragon's table rather than a figure Paragon states.

Is CELPIP Speaking harder than IELTS Speaking?

Different rather than harder, and the differences run in both directions. CELPIP removes the human examiner, which helps people who freeze in interviews and hurts people who feed off a listener's reactions. It also asks you to describe a static image and then predict what happens next inside that same image, and to talk your way out of an awkward social situation, none of which IELTS sets as a discrete task. Preparation built purely on IELTS material leaves those four task types unrehearsed.

Which CELPIP test do I need, General or General LS?

That depends on your application, and the official Who Accepts CELPIP page is the place to check rather than a blog. Structurally, CELPIP-General covers all four skills while CELPIP-General LS covers Listening and Speaking only, using the identical eight speaking tasks. Because speaking is one of only two scored bands on LS, it carries half your result while taking about a fifth of the sitting, so a weak speaking performance has nowhere to hide, which is visible in its lower mean score.