The most-cited number about CELPIP Speaking is wrong. Task 1 gives you 90 seconds to speak, not 60, and a page ranking in the top three for the Task 1 query says “you will have 60 seconds to record your spoken answer”. Google’s summary splits the difference and invents a range, “60 to 90 seconds, depending on the exact prompt layout,” as though the timing varied by prompt. It does not. Each task has a fixed published figure.
Those figures are in one table in the test owner’s own study pack. Here it is, with the two tasks the corpus consistently gets wrong flagged.
The Eight Tasks With Their Published Times
From the CELPIP Speaking Pro: Target 9+ Study Pack, page 5, read on 12 November 2026:
| Task | Preparation | Speaking |
|---|---|---|
| 1. Giving Advice | 30 seconds | 90 seconds |
| 2. Talking about a Personal Experience | 30 seconds | 60 seconds |
| 3. Describing a Scene | 30 seconds | 60 seconds |
| 4. Making Predictions | 30 seconds | 60 seconds |
| 5. Comparing and Persuading | Part 1: 60 seconds, Part 2: 60 seconds | Part 1: none, Part 2: 60 seconds |
| 6. Dealing with a Difficult Situation | 60 seconds | 60 seconds |
| 7. Expressing Opinions | 30 seconds | 90 seconds |
| 8. Describing an Unusual Situation | 30 seconds | 60 seconds |
Add the speaking column and you get 540 seconds, exactly the nine minutes of recorded speech inside a component the pack describes as taking “about 15 minutes to complete.” Add the preparation column and you get 360 seconds. The arithmetic closes, which is how you know the table is complete rather than approximate.
Three readings worth taking from it.
Tasks 1 and 7 are the long ones. Ninety seconds of advice and ninety seconds of opinion, against sixty everywhere else. A candidate who has rehearsed a sixty-second shape will run dry with thirty seconds of clock left, twice, on the two tasks that carry the most developed argument.
Task 6 doubles your preparation. Sixty seconds instead of thirty, because the task is a difficult interpersonal situation and the published design gives you time to decide on a position before you take it.
Task 5 is not one task. Part 1 gives sixty seconds of preparation and no speaking at all. Part 2 gives another sixty of preparation and sixty of speaking. So it carries 120 seconds of total preparation, more than any other task, and it is the only part of the component where you prepare and then do not speak. Every page that lists Task 5 as a single “30 seconds prep, 60 seconds speaking” row has flattened a two-part task.
Two On-Screen Bars Tell You Everything While You Speak
The official overview pack describes what the screen shows during a response, and both indicators matter more than candidates expect.
“When you begin speaking, a blue progress bar in the centre of the screen will show you how much time you’ve got left. Once the bar is completely blue, your time will be up.”
“As you speak, the grey bar behind the microphone icon will move up and down. It’s important that the grey bar not fall too low or the recording may not be audible.”
So the clock is visible and so is your volume. The same pack confirms that “responses are recorded and rated later,” and before the test begins you are “given an opportunity to test the recording device and hear how” you sound. The published advice is to “speak at your normal volume. Avoid speaking too loudly or too softly.”
The Screen Advances Itself And Recording Starts Without You
The same page states the mechanics plainly:
“For each task, when the preparation time is up, the screen will automatically move forward. You will hear a prompt (‘Start speaking now’) and recording will automatically begin.”
So there is no button to press when you are ready, and no way to buy an extra five seconds of thinking. Preparation time ends, the prompt plays, the recording runs. Candidates who plan to start by repeating the question back to themselves are spending recorded seconds on words that earn nothing.
Nothing published states that you may re-record or return to a finished task, and we are not going to assert the negative from silence. What is published is the automatic advance, which leaves little room for either.
Three To Five Raters Score Your Speaking, Not An Examiner
Several ranking pages describe “certified examiners” assessing your responses, and one tells readers what “examiners hear” all day. The word examiner does not appear in the test owner’s own description of how Speaking is scored. What appears on the results page is this:
“Each CELPIP speaking performance is rated by three to five speaking raters, and each CELPIP writing performance is rated by four to six writing raters.”
They work independently, with no knowledge of one another’s ratings. Tests are randomly assigned by an online system, test taker anonymity is maintained throughout, and if ratings disagree a benchmark rater is assigned automatically. Raters must be native speakers of English or at Canadian Language Benchmark 11 or 12, hold at least an undergraduate degree, and carry a TESL Canada-recognised certification or graduate training in language education or linguistics plus three years of teaching.
Not one page ranking on the first page of the five main Speaking queries states the three-to-five figure. It changes how you should think about the component. You are not performing for one person whose preferences you might read; you are producing a sample that three to five strangers will rate separately against published criteria.
That is also the reason there is no live examiner to interrupt, prompt or rescue a stalled answer. You speak into a microphone at a screen, in what the test owner’s guidebook calls a test “taken in a single 3-hour sitting with no separate Speaking test.”
Listenability Is The Dimension That Has No Writing Equivalent
Speaking is scored on four dimensions, named on the results page with their sub-factors:
| Dimension | What sits inside it |
|---|---|
| Content/Coherence | Number of ideas, quality of ideas, organization of ideas, examples and supporting details |
| Vocabulary | Word choice, precision and accuracy, range of words and phrases, suitable use of words and phrases |
| Listenability | Rhythm, pronunciation and intonation; pauses, interjections and self-correction; grammar and sentence structure; variety of sentence structure |
| Task Fulfillment | Relevance, completeness, tone, length |
Three of those four match the CELPIP Writing dimensions exactly. The fourth does not: Writing has Readability, Speaking has Listenability, and the swap is where most of the useful difference lives.
Look at what Listenability pools together. Pronunciation sits in the same dimension as grammar, and so do pauses, interjections and self-correction. That means a candidate who corrects themselves mid-sentence three times is spending the same dimension that their sentence variety is drawing on. The instinct to fix a small error out loud is expensive here in a way it is not in writing, where you simply delete the word.
As with Writing, each dimension “is divided into five performance levels” and the component score is “derived from the dimensional ratings assigned by the raters,” then transformed into a level by standard-setting rules. There is no published 0-to-5 scale and no published average.
What The Speaking Levels Ask For, From 3 To 12
The test owner’s level descriptor chart dated May 2026 carries Speaking on page one, each level with a label and a set of “At this level, I can” statements.
The boundary that matters for permanent residence sits between 8 and 9. At level 8 a candidate can “present concrete and some abstract information that is supported with moderately complex reasons” and “speak with some control of complex grammatical structures.” At level 9 the wording shifts to “use common and context-specific vocabulary,” “use some figures of speech and idioms,” and “adjust my speaking style and tone to a range of audiences and situations.”
So the jump from 8 to 9 is not fluency. It is register and idiom: speaking differently to a colleague than to a stranger, and using the fixed phrases a fluent speaker reaches for. The chart’s own level 9 performance profile puts it this way: mistakes in grammar, word choice and pronunciation “rarely make it hard for people to understand my meaning.” Errors are permitted at 9. Flat, undifferentiated register is not.
Levels 11 and 12 ask for the same thing more precisely, and only level 12 requires “a very broad range” of structures and “consistently fluent” delivery. Our score chart page carries all ten labels and the conversion to Canadian Language Benchmarks.
Speaking Band 8 Catches A Quarter Of All Candidates
The test owner’s 2025 test taker data report publishes the share of CELPIP-General candidates at each Speaking band:
| Band | Share of candidates |
|---|---|
| 12 | 4.1% |
| 11 | 8.2% |
| 10 | 8.0% |
| 9 | 14.6% |
| 8 | 24.4% |
| 7 | 18.3% |
| 6 | 12.6% |
| 5 | 5.9% |
| 4 | 2.6% |
| 3 | 0.9% |
| 0 to 2 | 0.3% |
34.9% of candidates reach Speaking 9 or above, against 24.1% in Writing. Speaking is the easier of the two productive skills to lift to Canadian Language Benchmark 9, by roughly eleven percentage points of the candidate population.
India’s Speaking mean is 8.03 against a global 7.94, and India’s mean beats the global figure on all four skills. The component mean overall is 7.94 with a standard deviation of 1.96, so the distribution is wider than Writing’s. There is more movement available here, in both directions.
The report publishes no sample size and no reporting dates beyond the year, so read the percentages as shape.
Preparation Seconds Are The Thing To Rehearse
Everything above points at one practical conclusion that no amount of vocabulary work substitutes for. Eight tasks, with 30 or 60 seconds of preparation each, screens that advance by themselves, and 90-second answers on tasks 1 and 7. The constraint is not your English. It is deciding what to say inside half a minute and then filling the full window without self-correcting your way through Listenability.
The published study pack says as much: “Practice giving yourself longer than the 30-60 seconds of prep time to come up with main ideas and details before recording a response,” then compress. Our CELPIP coaching runs the eight tasks at their real clock, and if you have not yet settled the format, what CELPIP is is the starting point. Candidates weighing this component against an IELTS speaking interview should read CELPIP or IELTS for Canada PR, where the two speaking formats diverge most.
Key Facts
- Speaking is about 15 minutes, and it is the last component of the test.
- Eight tasks, 540 seconds of recorded speech, and 360 seconds of preparation in total.
- Task 1 gives 90 seconds of speaking, not 60. So does Task 7. The corpus routinely says 60.
- Task 6 gives 60 seconds of preparation, double the usual 30.
- Task 5 has two parts. Part 1 has preparation but no speaking; Part 2 has both. It carries 120 seconds of preparation.
- The screen advances automatically and recording begins on the prompt “Start speaking now.”
- Three to five raters score each speaking performance, independently and anonymously, with a benchmark rater on disagreement. There is no live examiner.
- The four dimensions are Content/Coherence, Vocabulary, Listenability and Task Fulfillment. Listenability pools pronunciation with grammar, pauses and self-correction.
- 34.9% of candidates reach Speaking 9 or above, against 24.1% in Writing.
- India’s Speaking mean is 8.03, above the global mean of 7.94.
Questions Candidates Ask Before Booking
How many tasks are in CELPIP Speaking and how long is it?
Eight tasks, taking about 15 minutes in total. They add up to 540 seconds of recorded speech and 360 seconds of preparation. Speaking is the last component of the CELPIP Test.
How long do you get to speak in CELPIP Speaking Task 1?
90 seconds, after 30 seconds of preparation. This is the single most misquoted figure about the component: two pages ranking in the top three for the Task 1 query say 60 seconds, and Google’s summary gives a range of 60 to 90 seconds as though it varied by prompt. The published figure is a flat 90.
What are the eight CELPIP Speaking tasks?
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, in that order.
How much preparation time does CELPIP Speaking Task 6 give you?
60 seconds, double the 30 seconds given on tasks 1, 2, 3, 4, 7 and 8. Task 6 is Dealing with a Difficult Situation, and the extra preparation is published design rather than an error. Google’s summary states 30 seconds for this task.
Why does CELPIP Speaking Task 5 have two parts?
Because the task compares two options and then persuades. Part 1 gives 60 seconds of preparation and no speaking at all; Part 2 gives another 60 seconds of preparation and 60 seconds of speaking. That makes 120 seconds of total preparation, the most of any task, and it is the only point in the component where preparation time is followed by no recording.
Is CELPIP Speaking marked by a computer or a human?
By humans. Each speaking performance is rated by three to five speaking raters who work independently, with no knowledge of each other’s ratings, on randomly assigned and anonymised tests. A benchmark rater is assigned automatically if the ratings disagree. Raters must be native English speakers or at CLB 11 or 12, hold at least an undergraduate degree, and carry recognised teaching credentials.
Is there a live examiner for CELPIP Speaking?
No. You speak into a microphone at a computer in the same sitting as the other three components, which the test owner’s guidebook describes as a single sitting with no separate Speaking test. Pages describing “certified examiners” assessing your answers are using a word the test owner does not use for this component.
What are the CELPIP Speaking scoring criteria?
Content/Coherence, Vocabulary, Listenability and Task Fulfillment. Three of the four match the Writing dimensions; Listenability replaces Readability. Listenability covers rhythm, pronunciation and intonation, pauses, interjections and self-correction, grammar and sentence structure, and variety of sentence structure.
How hard is it to get 9 in CELPIP Speaking?
The publisher’s 2025 data shows 34.9% of CELPIP-General candidates reaching Speaking 9 or above, with 24.4% landing on exactly 8. The published level 9 descriptors ask for context-specific vocabulary, some idioms, and adjusting style and tone to a range of audiences. Errors are tolerated at level 9 provided they rarely obscure meaning, so flat register costs more than imperfect grammar.
Can you re-record an answer in CELPIP Speaking?
Nothing published says you can. When preparation time ends the screen advances by itself, the prompt “Start speaking now” plays, and recording begins automatically, which leaves no point at which a re-record would be offered. The test owner does not publish an explicit statement either way, so treat the automatic advance as the operative fact.

