CELPIP Speaking Phrases — The Complete Bank for All 8 Tasks

CELPIP Speaking has eight tasks, all recorded into a microphone with no live examiner, in Canadian English, with short prep and speaking windows (most tasks give 30 seconds to prepare and 60 or 90 seconds to speak). Under that pressure, the test-takers who score highest are rarely the ones with the cleverest ideas — they are the ones who can open, organise and connect their answer without freezing. This page is a reference bank: organised tables of natural, ready-to-use spoken phrases you can memorise and slot into any answer. Skim it, pick a handful per category, say them aloud until they feel automatic, and you will sound far more fluent on test day.

What the examiner is looking for

CELPIP Speaking is rated on four criteria. Knowing what each one rewards tells you why these phrases work.

Listenability is the criterion most candidates underestimate. The raters are listening for a steady, confident delivery — not perfection. A natural "Well, let me think for a second…" sounds far better than three seconds of silence. The phrases here exist to protect your flow.

How to use this bank effectively

Do not try to learn everything. Choose two or three phrases from each category and make them your defaults. Practise them out loud, not just by reading — your mouth needs the muscle memory. Rotate your openers so you are not robotic, and always prioritise sounding natural over sounding fancy. A memorised phrase delivered stiffly hurts Listenability; the same phrase said relaxed and on-rhythm lifts your whole answer.

Opening — starting your answer

The first five seconds set the tone. Begin immediately, restate the situation in your own words, and you buy yourself thinking time while sounding in control. Suits all tasks.

Giving an opinion and taking a stance

The raters want a position they can follow. Commit early and clearly, then defend it. Strongly suits Tasks 5, 6, 7 and 8.

Giving reasons and examples

A strong answer is reasons plus evidence. Name how many reasons you have, then deliver them — it makes your answer feel organised and complete. Suits every opinion or advice task.

Comparing options (Task 5)

Task 5 asks you to choose between two options and persuade someone. Weigh them openly, then commit. These phrases keep the comparison balanced and clear.

Making predictions (Task 4)

Task 4 asks you to predict what will happen next in a scene. Use future and likelihood language confidently, with reasons drawn from the picture.

Describing scenes and objects (Tasks 3 and 8)

Task 3 is describing a picture so a listener who cannot see it understands; Task 8 is describing an unusual situation or object from an image. Move logically — foreground to background, left to right — and use position language.

Telling a personal story (Task 2)

Task 2 asks you to describe a personal experience. Use clear past-tense narration and time markers so the listener follows the sequence.

Being tactful in difficult situations (Task 6)

Task 6 puts you in an awkward situation where you must deal with someone diplomatically. Soften your language, acknowledge feelings, and offer a way forward.

Buying time and natural fillers

This is the single most useful bank for Listenability. When your mind goes blank, a natural filler keeps you talking and buys a second to think. Suits every task — but use sparingly, not as a verbal tic.

Concluding and wrapping up

A clean ending signals control and helps you finish before the timer cuts you off. Land your conclusion around the 50- or 80-second mark. Suits all tasks.

The fill-in template

For any opinion or advice task, this skeleton turns the banks above into a complete 60-second answer. Drop in your chosen phrases.

[Opener — restate the situation in your own words]
[Stance — "I strongly believe that ___" / "If it were up to me, I'd ___"]
[Reason 1 — "The main reason is that ___"]
[Example — "For instance, ___" / "Take my own experience — ___"]
[Reason 2 — "On top of that, ___"]
[Filler/transition if needed — "The thing is, ___"]
[Conclusion — "So all in all, I'd really recommend ___"]

Model answer

Below is a Task 5 response (choosing between two options) built entirely from the banks above, sounding natural rather than scripted.

Why this answer scores high

• Opener and stance (first lines): "Honestly, if it were up to me, I'd go with…" commits to a position in the first five seconds — strong Content & Coherence and Task Fulfilment.
• Comparison language: "Both options have their upsides, but…" and "Sure, the downside is X, but it's worth it because…" show the rater a genuine weighing of choices, which Task 5 specifically rewards.
• Reasons plus example: "The main reason is…", "On top of that…" and "take my own experience" give structure and concrete evidence — boosting Content and Vocabulary.
• Listenability: contractions, the filler "Honestly," and a steady rhythm make it sound like real speech, not a recital.
• Clean conclusion: "At the end of the day…" and "So all in all, I'd really recommend…" land the answer with time to spare.

What CLB 7, 9 and 11 look like

• CLB 7: Uses basic openers and connectors ("I think", "because", "and") and addresses the task, but reasons are thin, vocabulary repeats, and there are noticeable pauses while searching for words.
• CLB 9: Deploys varied phrases from several banks, develops two clear reasons with an example, controls pace, and uses fillers naturally to avoid dead air. Minor slips do not break the flow.
• CLB 11: Phrases sound effortless and idiomatic, the structure is invisible because it is so smooth, vocabulary is precise and flexible, and delivery has natural stress and intonation throughout. Concessions and persuasion feel genuinely spontaneous.

Common mistakes to avoid

• Sounding scripted. Memorised phrases delivered stiffly hurt Listenability. Practise out loud until they feel like your own words.
• Overusing fillers. One "Well,…" buys time; five "you know"s in a row is a verbal tic. Rotate them.
• Forgetting the actual task. A beautiful comparison answer fails Task 4 if the task asked you to predict. Match the bank to the prompt.
• No stance. On Tasks 5, 7 and 8, sitting on the fence kills Content & Coherence. Commit to one side.
• Running out of time mid-sentence. Aim to begin your conclusion around 50 seconds (60-second tasks) or 80 seconds (90-second tasks).
• Long silent pauses. Silence reads as a breakdown in fluency. A filler phrase is always better than dead air.

Buying time and staying fluent

If you blank completely, chain a couple of these to recover without panicking: "That's a really good question, let me think for a second… so, the way I see it…". The rater hears a fluent speaker buying a moment, not someone stuck. Keep one go-to recovery phrase ready for every task and you will never face true silence on the recording.