CELPIP Listening Part 5: Discussion

Track multiple speakers in a discussion.

~8 min each
0 of 29 sets done

Practice sets

Up next in Listening Part 6 · Listening to Viewpoints Preview
Guide

Listening Part 5 is the only part of the CELPIP test with video. Three people discuss something and disagree, and the questions are built almost entirely around who thought what.

What is CELPIP Listening Part 5?

Three speakers, on video. This is officially the only video clip in the Listening test: every other part's specification says "one audio clip", and Part 5 alone says "one video clip".

QuestionsTimeMedia
8~9 minutesOne video clip, 1.5 to 2 min

Register is relatively informal, and it mixes facts, opinions and emotions. The speakers will sometimes disagree. Official examples: teammates deciding where to hold a tournament, neighbors planning a block party, classmates dividing up a group project.

On the question count. Some prep sites say Part 5 has five questions. It has eight. Both official sources say so, and the arithmetic proves it: 8 + 5 + 6 + 5 + 8 + 6 = 38, which is exactly the published Listening scale. Five would make the section total 35, and the official /38 score would be impossible.

CELPIP Listening Part 5 format: one screen, pooled timer, any order

This part uses CELPIP's second Listening interface, and it works nothing like Parts 1 to 3. The questions are sentence completions that you read on screen, they all appear together on one screen after the audio, and they share a single pooled time block rather than a 30-second-per-question timer. You may answer them in any order.

That last point is a real, exploitable advantage: sweep the questions your notes clearly answer, bank those points, and spend what is left on the two or three that need reconstruction. You cannot do that in Parts 1 to 3.

How CELPIP Listening Part 5 is scored

Computer-scored, one point each, no penalty for guessing, from the section's 38 scored questions. At eight questions, Part 5 is tied with Part 1 as the largest block in the Listening test, but it sits much further up the difficulty ramp.

The video plays once, starts automatically, and cannot be paused or replayed.

CELPIP Listening Part 5 strategies

  1. Build a three-column speaker grid before the clip, keyed to on-screen position. This is a direct official instruction: "label them according to the speakers' positions on screen as they are introduced." Use L, C, R and backfill names later. Names fly past once; screen position never changes. This is the single highest-leverage habit in Part 5.
  2. Log a stance per speaker, not a transcript. Questions read like "Dr. Jenkins would most likely agree with Larry that ___", which is unanswerable from content notes. It needs an attribution map.
  3. Mark the mind-changers with an arrow: R: against → OK if budget. Speakers characteristically move: someone opens sceptical and gets talked round. Do not overwrite their opening position, record the shift.
  4. Use faces only for the inference questions. Official guidance says inference questions probe feelings communicated indirectly through tone of voice and body language.
  5. Exploit the "any order" rule. Sweep the specific-information questions your notes clearly answer, bank them, then spend what is left on the two or three inference items.

Common CELPIP Listening Part 5 mistakes

  • Attribution errors: the defining Part 5 failure. Three speakers, names stated once, and questions built around who held which view. People understand the discussion perfectly and still lose the points.
  • Watching it like television. The video invites passive viewing, and Paragon cared enough about this to warn in writing that you will not be tested on irrelevant visual details. Meanwhile the pen stops moving.
  • Freezing a speaker's opening position and missing the reversal or the condition ("I could support that, if we move the date").
Tip: Label your three columns L, C and R before the video starts, not by name. Names are said once and fly past. Where someone is standing on screen never changes, and you can fill in the names later, once the questions tell you which ones matter.

CELPIP Listening Part 5: frequently asked questions

Is CELPIP Listening Part 5 the one with the video?
Yes, and it is the only one. Paragon's specification for Part 5 says "one video clip (about 1.5 to 2 minutes)"; every other part of the Listening test is audio only.
Does CELPIP Listening Part 5 have 5 questions or 8?
Eight. Both official sources say eight, and the arithmetic settles it: 8 + 5 + 6 + 5 + 8 + 6 = 38, which matches the official /38 Listening scale exactly. Some prep sites say five; five would make the section total 35, which is impossible.
Do I need to remember visual details from the video?
No, and Paragon says so explicitly: you will not be tested on visual details unrelated to the conversation, such as whether a character wore glasses or what color someone's shirt was. Watch faces for attitude, not for set dressing.
Can I pause or replay the video?
No. It plays once, starts automatically, and cannot be paused or rewound.
Should I watch the screen or look down at my notes?
Both, deliberately. Glance up at speaker transitions to lock in who is who and to read tone, and keep the pen moving the rest of the time. The inference questions are the only ones that need the faces.
Can I replay the audio?
No. Paragon is explicit: all audio clips are played one time only, they begin automatically, and they cannot be paused or rewound. This is the single biggest difference from most practice you will have done, and it is why note-taking, not comprehension, is usually the thing that needs training.
Can I take notes during CELPIP Listening?
Yes. You are given pen and paper at the test center and you can ask for more. Your notes are never collected or scored. (Some prep sites claim you get an erasable noteboard and marker. Paragon says pen and paper.)
Is there a penalty for a wrong answer?
No. Paragon states that points are not deducted for incorrect answers, so leaving a question blank is strictly worse than guessing. Always put something down before the timer runs out.