Part 5 is the only video task in the test: three people negotiate an everyday decision for about 2 minutes, then 8 drop-down questions appear on one screen with roughly 30 seconds each. Stems name speakers by name and by position on screen, and the answers live in who wants what, who objects, and who changes their mind.
Part 5 is the only video task in the whole test: you watch three people discuss an everyday decision for about 2 minutes, then answer 8 questions from drop-down menus on a single screen. The clip plays once, the questions appear only after it ends, and the stems identify speakers by name and by position on screen. The demo above is the exact interface you get on test day.
Listening is scored by computer, one point per correct answer and no penalty for wrong ones. Part 5 contributes 8 of the 38 scored Listening questions, the biggest single block in the test, and most of them ask for inference rather than plain fact recall.
Eight chances, each simply right or wrong, with nothing deducted for guessing. Fill every drop-down before the shared timer expires, an empty menu is the only option that can never score.
Part 5 has no separate score, its points join the Listening total out of 38, which converts to a CELPIP Level from 1 to 12 with difficulty balanced across test versions. Eight questions in one part means this clip moves your Level more than any other.
Officially the test checks general meaning, specific information, and inference. In Part 5 the balance tilts hard toward inference, stems ask what a speaker sounds like, would agree with, or probably means.
One experimental part may be mixed in without counting, and it looks identical to the scored ones. Never write a part off as the fake, play every clip like it counts, because you cannot tell.
The screen introduces the scene and often the names, such as one woman and two men on a break at work. Draw a three column grid and label it by position as the speakers appear, left, middle, right, because stems say the man on the left as often as they use names.
For each speaker capture what they want and why, what they object to, and any condition they attach. Watch faces and tone too, attitude questions draw on them, but ignore shirt colours and props, visual details unrelated to the conversation are never tested.
In most discussions one speaker moves from objection to a conditional yes, listen for as long as and provided that, and the group lands on a final plan. Final positions beat opening positions, and both the concession and the final decision are near guaranteed questions.
All 8 stems share one countdown. Classify each one fast, whole clip, single fact, or inference, answer the confident ones first in any order, match meanings rather than exact words, and leave no menu empty when the screen moves on.
These are the official instruction patterns plus the signal lines that structure a three way discussion. The concession phrases are gold, they mark the stance change that questions circle around.
Three similar voices trade opinions fast, and distractors love to hand one speaker another's point. A labelled grid, one column per person, keyed to screen position, makes this trap collapse on sight.
A speaker who resists early and concedes later is the Part 5 signature. Options quoting the early objection against the final plan are traps, always answer from where the discussion landed.
Could, might, and leaning toward are tentative, and as long as attaches a condition. Options that upgrade a maybe into a definite plan, or quietly drop the condition, are wrong on purpose.
Part 5 is video because expressions and gestures carry attitude, and inference stems ask how a speaker feels. Keep your eyes on the screen and your notes short, but skip visual trivia, clothing and props are never tested.
Eight questions share roughly 4 minutes. A stubborn inference item can eat three easy points, so bank the sure answers first, return with fresh eyes, and guess rather than leave any menu blank.
Yes. Every other listening part is audio only, Part 5 plays one video of about 2 minutes. It starts automatically, plays one time only, and cannot be paused, and in exam mode here the single play is enforced just like test day.
Constantly. Official stems read like Peter, the man on the left, so notes keyed only to voices break down. Label your note columns left, middle, and right as the speakers first appear, then attach names once you catch them.
Let it go and stay with the discussion, there is no replay and no pause. Keep notes to two or three words per turn so your eyes stay on the speakers, the video itself, faces and tone, answers the feeling questions your pen cannot catch.
