CELPIP Listening Part 2 (Daily Life Conversation): The Easy One People Lose Marks On

What Part 2 actually is

Part 2 is Listening to a Daily Life Conversation: two speakers, one continuous clip of about 1.5–2 minutes, and five questions hanging off it.

Do the arithmetic and it is the least audio you get per question of any two-speaker conversation in the section — Part 1 gives you three sections of roughly 1–1.5 minutes each for its eight questions, Part 3 gives you 2–2.5 minutes for six. Part 2 gives you under two minutes for five.

And it is the friendliest audio on the test. Nobody is complaining about a defective toaster, nobody is explaining municipal recycling policy. Two people are arranging something.

That combination — easy language, high question density — is exactly what makes Part 2 expensive. Everyday conversation is low-information-density right up until a question asks which of two similar details was true, and then it turns out you were never storing details at all. Nothing in an ordinary conversation feels important while you are hearing it. That is what "ordinary" means.

The clip plays once. Officially: "All audio clips are played one time only." When it ends, the only version of it that still exists is the one in your head and the one on your paper.

The failure: listening like a guest

There is a specific way people lose these five marks, and it is not vocabulary and it is not speed.

You listen to Part 2 the way you would listen to two colleagues chatting near your desk — following the mood, the relationship, roughly what they landed on. Polite, passive, gist-level. Listening like a guest.

Then the questions arrive and they want the ledger. Who offered what. Which day was ruled out and which one survived. What the second speaker actually agreed to, as opposed to what they first suggested. Whether the forty dollars was a fee or a discount and whether it applied to Saturday or to Friday.

You knew all of that while the audio was playing. You just never wrote it down, because it did not feel like information at the time.

The tell is the feeling afterwards. After Part 1 people say "that was fast." After Part 2 people say "that was easy" — and then score four out of five, or three.

Worked example (original AlmiWorld practice item)

Nadia: Did the furniture people ever call you back? Marcus: This morning. They can do Thursday afternoon — sometime between two and five. Nadia: Thursday's bad for me. I've got the dentist at three. Marcus: I asked about Friday. Friday's completely full. They could do Saturday morning, but weekend deliveries have a fee — forty dollars. Nadia: Forty? No. Let's keep Thursday. Can you be here for it? Marcus: I finish at noon on Thursdays, so I'll be home from one. Nadia: Good. Oh — they said whoever placed the order has to sign for it. That's me. Marcus: Then get back from the dentist before five. Or call them and put my name on the order instead. Nadia: I'll call them. Easier than rushing back.

Question: What is Nadia going to do?

A. Sign for the delivery after her dentist appointment B. Phone the company and add Marcus's name to the order C. Move the delivery to Saturday morning D. Ask the dentist to see her in the afternoon

Answer: B.

Now the part that matters — why the other three are wrong, given that every one of them is built from something that genuinely happened in the audio.

A is the right detail attached to the wrong plan. Signing after the dentist was said out loud — by Marcus, as the first of two options, and Nadia turned it down in her last line. If you stored "sign" and "dentist" as loose facts rather than as a proposal that got rejected, A looks correct.

C is the right thing at the wrong time. Saturday morning was real and available. It died on the forty-dollar fee, one line later. Options like this survive on the fact that you heard the day clearly and stopped tracking what happened to it.

D is a swapped detail plus a sensible idea nobody had. "Afternoon" belongs to the delivery window — two to five — not to the dentist, who is at three and never moves. Moving a dentist appointment is what a reasonable person might do, which is precisely why it is offered to you. Part 2 rewards what was said, not what would be sensible.

Notice that none of the wrong options contains information that was absent from the audio. That is the standard shape of a Part 2 distractor: correct material, reattached to the wrong speaker, the wrong day, or the wrong stage of the conversation.

The shape that works

1. Decide before the clip starts that this is a two-column problem, not a story. One column per speaker. Everything you write goes under a name.
2. Write only the ledger items: days, times, numbers, prices, names, and decisions. Not adjectives, not reasons, not full sentences.
3. Mark reversals. When something is proposed and then killed, strike it through. A struck-out Saturday is worth more than a remembered one.
4. Track the last state of every arrangement, not the first. Part 2 conversations move — the answer is usually where the topic ended, not where it started.
5. Use the reading time on the options properly. Answer choices are not read aloud to you — the format page is explicit that you have to read them. For Parts 1–3 the guidance is around 30 seconds to hear and answer each question, so read every option against your notes rather than picking the first one that sounds familiar.
6. When two options both feel true, ask which speaker said it and when. That question resolves most Part 2 ties on its own.

What to write down when nothing is happening

Paper and pen are provided at the test centre. Here is what the conversation above should have looked like on it — about fifteen seconds of writing, spread across two minutes:

NADIA            |  MARCUS
dentist Thu 3    |  Thu pm 2-5  ✓
~~Sat am +$40~~  |  ~~Fri full~~
signs (ordered)  |  home from 1
→ WILL CALL: add M to order

Four lines and an arrow. The arrow line is the one that answers the question.

Things to deliberately not write: why Friday is full, how Nadia feels about forty dollars, the word "furniture". Anything you can reconstruct is wasted pen-time, and pen-time is attention you are not spending on the audio.

Signal phrases that mean a detail just changed owner or status — these are the moments to look up and write:

• Actually — / On second thought — / In that case —
• Let's just… / Never mind, / Forget it, let's…
• I was going to… but…
• Can you…? / I'll do it / You do it (attribution is about to matter)
• Instead / rather than / no, wait
• Any number said out loud, and any day of the week

What actually costs you the mark

• Relaxing because the language is easy. Easy language still carries hard facts.
• Taking notes in sentences. By the time you finish writing one, two more decisions have gone past.
• Recording what was proposed and not what was agreed.
• Losing track of which speaker owns a detail — the single most common Part 2 distractor.
• Choosing the option that sounds most reasonable in real life rather than the one that was said.
• Treating "I remember that clearly" as evidence. Distractors are built from things you remember clearly.
• Drifting during the last twenty seconds, which is where these conversations usually settle.

Remember that you cannot return to a previous part once you have moved on — Part 2 is finished when it is finished, and the test moves the screen forward on its own.

Try one

[Try Listening Part 2 →] — one full task, marked, no card required.