How to Self-Review Your CELPIP Speaking Recordings: A 4-Category Checklist

Recording your practice answers is the easy part. Listening back is where most people give up. They play the clip once, cringe at their own voice, think "that was okay, I guess," and record the next one. That isn't review. It's just listening. A real review uses the same categories CELPIP raters use, finds the weakest one, and tells you exactly what to change in your next recording. This post gives you that process as a checklist you can use on every recording.

Want a second opinion to check your self-scoring against? Get an evaluation of a Speaking response → celpipuni.com/evaluation.

TL;DR

Review each CELPIP Speaking recording against the four official Performance Standards categories, one at a time, in four separate listens: Content/Coherence, Vocabulary, Listenability and Task Fulfillment. Paragon lists specific factors for each. For Content/Coherence: number, quality and organization of ideas, plus examples and details. For Vocabulary: word choice, suitable use, range and precision. For Listenability: rhythm, pronunciation and intonation; pauses, interjections and self-correction; grammar; sentence variety. For Task Fulfillment: relevance, completeness, tone and length. Note one problem per category, pick the weakest category, and re-record the same prompt fixing only that. A transcript makes this much easier.

What are the four CELPIP Speaking rating categories?

Paragon's CELPIP Speaking Performance Standards rate every response on four dimensions. The table below lists the factors Paragon gives for each, along with the guiding question it pairs with that dimension.

Source: Paragon's official CELPIP Speaking Pro study pack. Paragon's score reports also include Level Descriptors and "things to work on" for your level, so your official report is worth rereading alongside this checklist.

These sixteen factors are your review checklist. Nothing else is officially rated.

How should you set up a self-review session?

1. Record under real conditions. Use the real prep and speaking times (30 or 60 seconds of prep, 60 or 90 seconds of speaking, depending on the task), out loud, with no restarts.
2. Wait before you listen. Even an hour helps. Listening straight away, you hear what you meant to say, not what you said.
3. Make a transcript. Type it yourself, or use a phone's speech-to-text and then correct it against the audio. Write fillers like "um" and "uh" exactly as you said them. Paragon's own study pack suggests making a transcript of a recorded response and spending 20–30 minutes editing it to make it stronger.
4. Listen four times, one category per listen. If you try to judge everything at once, you'll mostly notice grammar, because that's the easiest thing to hear, and miss the rest.

Checklist 1: How do you review Content/Coherence?

Listen once, transcript in hand, asking: how well are my ideas organized and developed?

• Number of ideas: right for the task? Two or three developed points usually fill 60–90 seconds. One idea repeated, or six undeveloped ones, are both problems.
• Quality of ideas: does each idea answer the prompt, or is it generic ("communication is very important")?
• Organization: could someone draw your structure from the transcript? Is there a clear opening, an order, and a closing?
• Examples and supporting details: does each main idea have something concrete, like a person, a place, a number or a consequence?

Quick test: underline every specific detail in your transcript. If a whole idea has no underlining, that's your Content/Coherence fix.

Checklist 2: How do you review Vocabulary?

Listen again, asking: what's my range, and do I use it naturally?

• Word choice: are words accurate for the meaning? Watch for near-misses ("I was very boring" instead of "bored").
• Suitable use: do phrases fit the situation and the listener? A formal phrase to a friend, or slang to a customer, is a suitability problem.
• Range: count repeats. If "good," "very," "important" or "thing" appear four or more times in one response, your range is narrower than you think.
• Precision and accuracy: do you use specific words ("the line was dragging on," "short-handed") or vague ones ("it was not nice")?

Quick test: circle every word you used three or more times. Replace two of them with something more precise in your re-record.

Checklist 3: How do you review Listenability?

This is the one category where you need the audio, not just the transcript. Ask: how easy is this to listen to and understand?

• Rhythm, pronunciation and intonation: does your voice rise and fall naturally, or stay flat? Are any words hard to make out?
• Pauses, interjections and self-correction: count fillers ("um," "uh," "like") and pauses of more than about three seconds. Note where they cluster. Clusters usually mean you ran out of plan, not out of English.
• Grammar and sentence structure: are there errors that make meaning unclear? Tense slips in Task 2 stories and future forms in Task 4 are common.
• Variety of sentence structure: do all sentences follow the same pattern? Look for at least a few complex sentences with "which," "although," "because," "if" or "once."

Quick test: mark each pause on the transcript with "/". Then look at what you said just before each one. A pattern there tells you what to change in your prep.

Accent is not a category. Paragon has said publicly that raters don't score accent; what matters is whether they can understand you. Focus on clarity and rhythm, not on sounding like someone else.

Checklist 4: How do you review Task Fulfillment?

Read the prompt again, then listen, asking: how well did I follow the instructions, and was my tone right?

• Relevance: did you answer this prompt? For example, predictions in Task 4 rather than description, or speaking to the person in Task 6.
• Completeness: did you do everything the prompt asked, such as giving advice and reasons, or choosing an option and comparing?
• Tone: does it match the relationship: friend, manager, customer, neighbour?
• Length: did you speak for most of the time you had? Write down your actual speaking time. Stopping well short is a length problem.

Quick test: write the prompt's instruction verb at the top of your transcript ("advise," "predict," "persuade," "explain your decision"). Then check whether every paragraph actually does that.

How do you turn a review into improvement?

This is where most self-review fails. You've found eight problems. Now what?

1. Give each category a rough level for this recording. Use the CELPIP 1–12 scale, or simply "strong / okay / weak." Being consistent matters more than being exact.
2. Pick the weakest category. Only that one.
3. Write one fix as an instruction. Not "improve vocabulary" but "replace 'good' and 'important' with precise words."
4. Re-record the same prompt right away, fixing only that. Same prompt, so the only thing that changes is the fix.
5. Compare the two recordings. Did the fix work without breaking something else?

Repeat over several recordings until the weakest category catches up, then move to the next one.

How do you know if your self-scoring is accurate?

Self-scores tend to be generous, because you know what you meant. Two ways to check:

• Compare against official samples. Paragon's Score Comparison Chart on celpip.ca has sample Speaking responses at each CELPIP level, with transcripts and analysis by dimension. Listen to the level you're aiming for, then to yours.
• Get an outside score now and then. A trained grader or a rubric-based evaluation tells you whether your "8" is really an 8. Check your self-scoring against an outside score every few weeks so it stays honest.

Your self-review checklist, in short

• Recorded under real timing, no restarts
• Waited before listening
• Transcript made, fillers included
• Four listens, one category each, using Paragon's sixteen factors
• One weakest category chosen, one fix written as an instruction
• Same prompt re-recorded with only that fix
• Self-scores checked against official samples or an outside evaluation regularly

Want to check your self-review against a rubric-based score? CelpipUni evaluations rate your response on all four categories → celpipuni.com/evaluation.

The real value of self-review is noticing patterns

One recording tells you very little. Ten reviewed recordings show you patterns: your pauses always come at the start of the second point; you say "very" in every task; your Task 6 tone is fine but your Task 1 tone is too formal. Patterns are what you can fix, because they point at a habit, not a one-off slip.

So keep your reviews in one place, like a notebook or a spreadsheet with the date, task, four category notes and the one fix. After two weeks, read the "weakest category" column from top to bottom. If the same category keeps coming up, that's your study priority for the next month.

Tools and resources

• Paragon's CELPIP Speaking Performance Standards and Score Comparison Chart on celpip.ca. These are the official categories and level-by-level samples.
• CELPIP Speaking scoring criteria explained: celpipuni.com/celpip-speaking-scoring-criteria
• How CelpipUni's AI feedback works: what an automated rubric check can and can't tell you: celpipuni.com/how-ai-feedback-works
• CELPIP Speaking practice: the daily 25-minute system: celpipuni.com/celpip-speaking-practice

Frequently asked questions

What are the four CELPIP Speaking scoring categories?

Content/Coherence, Vocabulary, Listenability and Task Fulfillment. Paragon lists specific factors for each, such as the order of your ideas, range of words, pauses and self-correction, and length.

How do I score my own CELPIP Speaking practice?

Listen four times, once per category, using Paragon's factors as a checklist. Give each category a rough level, find the weakest, and re-record the same prompt fixing only that. Check your self-scores regularly against official sample responses or an outside evaluation.

Should I transcribe my CELPIP Speaking recordings?

Yes, at least some of them. A transcript makes structure, repeated words and missing details easy to see. Paragon's own study pack suggests transcribing a recorded response and spending 20–30 minutes editing it to make it stronger.

Does my accent affect my CELPIP Speaking score?

Accent isn't a rating category. Paragon has said raters focus on whether they can understand you. Clear pronunciation, rhythm and intonation, which are part of Listenability, matter more than how "native" you sound.

How often should I self-review?

Review every recording briefly, and do a full four-listen review on at least one recording per practice session. Compare your self-scores with an outside evaluation every few weeks to keep them accurate.

What to do next

1. Today, record one Task 1 and one Task 7 answer under real timing.
2. Tomorrow, transcribe both and do the four-listen review.
3. Pick the weakest category across both, write one fix, and re-record.
4. Start a review log, and look back at it in two weeks.

Related reading

• CELPIP Speaking Scoring Criteria: The 4 Things Raters Score
• Why Your CELPIP Speaking Is Stuck at 7 (and How to Fix It)
• CELPIP Speaking Samples With Scores: Band 7 vs 9 Compared
• CELPIP Speaking Practice: The Daily 25-Minute System

The bottom line

Self-review only works when it uses the categories you're actually scored on. Record under real timing, transcribe, listen four times against Paragon's four dimensions and sixteen factors, fix only the weakest category, and re-record the same prompt. Keep a log so you can see your patterns, and check your self-scores against an outside score now and then.

Check your self-review against a rubric-based score → Get your Speaking evaluated.