Note-Taking for CELPIP Listening: What Not to Write

Every guide tells you to take notes. None of them tells you what to leave out — and leaving things out is the entire skill.

Paper and pen are provided at the test centre. The audio plays once:

All audio clips are played one time only.

So every word you write is a word you are not hearing. Writing more is not safer. Writing more is how people lose the thread.

The cost model: writing competes with listening

Notes are paid for out of the only budget that matters — attention while the clip is running.

You hear a price. You start writing "one hundred and forty dollars". By the time your pen reaches the second word, the speaker has said two more things, and one of them was the correction that changes the price. You now have a neat, legible, wrong number on your paper, and no memory of the sentence that made it wrong.

The audio does not come back. The test moves itself forward — "the test will automatically move forward to the next screen" — and you cannot return to a previous part. A note bought with three seconds of listening is not a note. It is a trade, and usually a bad one.

The failure: the transcription reflex

The failure has a name. It is the transcription reflex — writing down whatever you just heard because you just heard it, rather than because you will need it. It feels like diligence. What it produces is a beautiful record of the first thirty seconds of a two-minute clip and nothing after.

What is never worth writing

• Anything you will still know in ninety seconds. The topic. The setting. That one of them is annoyed. You are not going to forget the conversation was about a leaking pump.
• Anything the answer options will remind you of. You read the options on screen, and they hand you the nouns back. Store the relationships between those items instead.
• Complete words. "Thursday" is Th. "Because" is b/c. "Manager" is mgr.
• Full sentences. A sentence has grammar in it. Grammar is pure cost — nobody marks your paper.
• Anything you are writing because it feels productive. If you cannot name the question it might answer, put the pen down and listen.

What is always worth writing

One test: write it down if getting it backwards would be easy and expensive.

• Things that come in pairs and can be swapped — two names, two dates, two prices, two options. This is where marks actually go. If the clip contains Thursday and Saturday, your real risk is attaching the wrong fact to the wrong day.
• Numbers. Prices, times, quantities. Digits are cheap to write and impossible to reconstruct.
• Sequence. What has to happen before what.
• Who said which of two positions. In any clip with more than one voice, half the difficulty is attribution, not content.
• Anything corrected mid-conversation. People in these clips say "actually", "sorry, I meant", "no, wait". A correction is a question waiting to happen. Mark it and keep the old value visible.

A notation system small enough to remember

Keep it to about a dozen marks. A system nobody can recall under pressure is worse than none.

That is the whole system. Learn it in one sitting, then use only it — inventing a new symbol during a clip costs the same attention as writing a sentence.

The side-by-side

(original AlmiWorld practice item)

About forty-five seconds of a two-speaker exchange:

Dan: The technician can come Thursday morning — that's the fourteenth. You said you were away. Priya: I'm back Wednesday night. Thursday's fine. What's it going to cost? Dan: Ninety for the call-out, plus parts. He quoted one-forty for the pump. Priya: One-forty? Last month the other place said one-ninety. Dan: Sorry — one-ninety was the other place, yes. But hold on, his one-forty is the older model. The one we need is one-sixty. Priya: So around two-fifty all in. And if we wait until the sixteenth, the second technician is cheaper? Dan: Cheaper on the call-out. But he can't do Thursday or Friday.

What a good test-taker has on paper afterwards:

Th 14 = tech1 | 16 = tech2
P back Wed nite -> Th ✓
tech1: $90 call + pump
   140 ✗ old model  ! 160
   190 = OTHER place
   tot ~250
tech2: call ↓  ✗Th ✗Fri

Seven lines of nothing, and every trap in the extract is on it.

The other person's page for the same forty-five seconds:

The technician can come on Thursday morning
which is the fourteenth, but Priya said she
was away that week. However she is back on
Wednesday night so Thursday works fine for
her. The cost is ninety dollars for the call
out plus parts, and he quoted one hundred
and forty dollars for the pump.

Longer, tidier, much worse. Trace what it cost.

While that hand was writing however she is back on Wednesday night so Thursday works fine for her, the speakers were doing the one-ninety mix-up — so the page cannot say which quote belongs to which company. While it spelled out one hundred and forty dollars for the pump, Dan corrected himself to one-sixty. The number on that paper is wrong, and it is written in confident full sentences, which is exactly what makes it convincing thirty seconds later.

Then the notes stop. The second technician, the sixteenth, the cheaper call-out, the two days he cannot do — none of it exists on that page, because the pen was still on sentence two.

Where the notes go on the page, part by part

The audio composition is published, so set your paper up before each clip starts. Lengths are approximate.

• Part 1, Listening to Problem Solving — 2 speakers, three sections of about 1–1.5 minutes each: two columns, with a line ruled across between sections.
• Part 2, Listening to a Daily Life Conversation — 2 speakers, about 1.5–2 minutes: two columns.
• Part 3, Listening for Information — 2 speakers, about 2–2.5 minutes: two columns, and leave room — this is the longest of the two-speaker clips.
• Part 4, Listening to a News Item — 1 speaker, about 1.5 minutes: no columns, one running list, mostly numbers and names.
• Part 5, Listening to a Discussion — 3 speakers, a video clip, about 1.5–2 minutes: three columns, ruled before it plays.
• Part 6, Listening to Viewpoints — 1 speaker, about 3 minutes, the longest clip in the section (other official documents call it "Listening for Viewpoints"): one timeline down the page, because the position moves as the speaker goes.

Your eyes leave the audio — that's the point

You do not hear the answer options. You read them:

Test takers will be required to read answer choices in the Listening Test.

At some moment your eyes go to the screen and your ear stops doing the work. In Parts 1 to 3 that moment comes on a clock — those parts allow 30 seconds to hear and answer each question.

Your notes exist to survive that moment. Ask of every mark you make: will this still mean something to me when I am looking at four plausible options and not at the audio? If the answer is no, it was decoration.

What actually costs you the mark

• Writing full words while the correction goes past.
• Recording the topic — which you cannot forget — instead of the pairs, which you can.
• No speaker labels in a three-voice discussion.
• A symbol system invented mid-clip.
• Notes so long you have to re-read them, with 38 questions moving forward without you.

Try one

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