CELPIP wiki

CELPIP Speaking Task 4 Making Predictions Guide

Task 4 is stronger when you commit to what will probably happen next and support each prediction with a reason from the scene, instead of describing the picture again.

Quick answer

Predict what happens next, and give each prediction a reason.

Short answer: In CELPIP Speaking Task 4, say what will probably happen next in the scene: pick the main action, predict its outcome with future forms such as going to and will probably, and give a short reason for each prediction. Two or three linked predictions with reasons score better than a list of unsupported guesses.

Task 4 follows the scene-description task, and that is the trap: the score weakens when the answer keeps describing what is already visible instead of moving the scene forward.

Structure

A practical Task 4 structure protects clarity under timing

1. Anchor on the action

Name the main thing happening now in one line.

2. Main prediction

Say what will probably happen next to that action.

3. Give the reason

Point to the detail in the scene that makes it likely.

4. Second outcome and close

Add one linked prediction, then close on the most likely result.

Sample outline: current action in one line -> main prediction + reason -> linked second prediction -> likely-result close.

Timing and scoring

Committed predictions with reasons beat cautious lists

Prediction answers often collapse into hedging: maybe this, maybe that, nothing claimed. Raters hear more range and better coherence when you commit to a likely outcome and connect it to evidence in the scene.

Scoring relevance

Future forms, cause language like because and so, and a visible link between prediction and evidence all show grammar range doing real work.

What to review after recording

Did the answer move past description within the first sentence? Did every prediction come with a reason? Did the close commit to one outcome?

Common mistakes

Task 4 answers usually weaken in the same three ways

Mistake

Re-describing the scene

The answer repeats what the picture shows, which is the previous task, not this one.

Mistake

Reasons missing

Predictions arrive as a list of guesses with nothing connecting them to the scene.

Mistake

Hedging everything

Every line is softened with maybe until the answer never actually predicts anything.

What to do next

Use one recording to check the structure, then fix one thing

Need better speaking support?

Open the speaking target-9 page if the structure is clear but the delivery still sounds thin or hesitant.

Open target 9+

Need wider speaking diagnosis?

Use the speaking score guide if Task 4 is part of a broader task-fit or structure problem.

Open speaking score guide

FAQ

Task 4 questions

How is Task 4 different from Task 3?

Task 3 describes what the scene looks like now; Task 4 moves it forward and predicts what will probably happen next.

What grammar does Task 4 reward?

Future forms such as going to, will, and will probably, plus cause-and-effect language like because and so.

Do wrong predictions lose points?

No. Raters score the language and coherence of the prediction, not whether your guess turns out to be correct.

Related guides

Keep the Task 4 prediction loop moving