Predicting from Context
The Celpip Plus team
Written and reviewed against the official CELPIP scoring criteria
What you'll learn: how to predict what the questions will target when you can't see them in advance, using the part you're in and the first few lines of the recording.
You won't see the questions first
In CELPIP Listening the questions appear only after the audio ends. There's no window to read them ahead and no chance to listen out for a detail you know is coming. That doesn't make prediction useless. It moves it. Instead of predicting from question stems, you predict from two things you always have: the type of part you're in, and the opening lines.
Good prediction here isn't guessing the answer. It's deciding, within the first ten seconds, which kinds of detail are worth a line in your notes.
Prediction source 1: the part you're in
Each part has a shape, and the shape tells you what the questions tend to target. Know these before test day so they cost you no thinking time:
| Part | What to expect | What to note |
|---|---|---|
| 1 — Problem Solving | Two people solving a problem, in three segments | The problem, each proposal, the final agreement, any "if" |
| 2 — Daily Life Conversation | A casual exchange about a shared plan or topic | The original plan, every revision, indirect opinions |
| 3 — Information | Practical facts delivered quickly | Times, places, costs, requirements, and any changes |
| 4 — News Item | A short report on one story | Who, what, when, where, why, and what happens next |
| 5 — Discussion | A video of three people discussing a topic | Each speaker's stance, reasons, and any change of mind |
| 6 — Viewpoints | One speaker reporting several groups' views | Which group holds which view, and each view's reasons |
Prediction source 2: the opening lines
The first sentence or two almost always establish who is talking, where they are, and what the talk is about. That's enough to set up your notes.
- Who and where tell you the likely topics. A customer and a hotel clerk will talk about dates, rooms and prices. Two parents will talk about times, pickups and who does what.
- The first problem or topic tells you what the rest will be about. If the first line is a complaint, expect a proposal, an objection and a final fix. If it's an announcement, expect a list of facts.
- The first fact that changes ("it was going to be Tuesday, but...") tells you this recording deals in corrections. Write both values every time.
In Part 1 you get a third source: the earlier segments. Once you've answered segment 1's questions, you know the problem and the first idea on the table, so you can expect segment 2 to bring an objection, a complication or a new option.
Worked example
Part 1, segment 1 opens: "Hi, I booked the community room for my daughter's party on Saturday, but the confirmation email says Sunday."
Before the second sentence, you already know:
- The problem: a booking on the wrong day.
- The likely questions: what the mistake was, what the staff member offers, what the caller decides, and any condition attached.
- The likely traps: a first offer that gets rejected, and a day or time that changes.
So you set up your notes like this and fill them in as the conversation runs:
prob: booked Sat, email says Sun
offer 1:
offer 2:
agreed:
IF:
When the questions appear, you aren't searching a page of loose notes. Most of them point at a line you already labelled.
Use the answer choices after the audio
When the questions do appear, they come with their answer choices. Read the choices for what they have in common before you pick. If three choices are all times ("2:00," "2:30," "3:00"), the question is testing a specific time, probably one that was corrected, so check your notes for an arrow. If the choices are near-synonyms ("frustrated," "disappointed," "annoyed"), the question wants a precise word, so look for what the speaker actually said rather than your sense of their mood.
Prediction checklist
- I know the shape of all six parts well enough to set up my notes without thinking.
- I use the first one or two sentences to decide who, where and what before writing anything else.
- I write both values whenever a detail changes, since a change signals a likely question.
- In Part 1, I use what I learned in earlier segments to expect what the next one will bring.
- When the questions appear, I read the answer choices for what they share before choosing.
Common mistakes
- Waiting to "see what they ask" before taking notes. By the time the questions appear, the audio is over.
- Treating every sentence as equally worth noting instead of using the part's shape to decide what matters.
- Missing the opening lines while settling in, then spending the recording working out who is talking.
- Predicting the answer instead of the kind of detail to note, then hearing what you expected instead of what was said.
Recap
Without the questions in front of you, prediction comes from context: the shape of the part you're in, the opening lines, and in Part 1 the segments you've already heard. Use them to decide what goes in your notes within the first few seconds, and use the answer choices after the audio to see exactly which detail each question is testing.