CELPIP Speaking Task 4: Making Predictions – Template, Timing & Level 9 Strategy

CELPIP Speaking Task 4 is Making Predictions. You see the same type of illustrated scene used for picture-based speaking and explain what you think will happen next. The current 2026 CELPIP Speaking Pro material gives you 30 seconds to prepare and 60 seconds to speak.

Official CELPIP guidance emphasizes realistic predictions connected to what people are already doing. Your imagination helps, but the image is evidence: the best predictions grow naturally from visible actions, nearby people, and surrounding objects.

Quick strategy: choose three visible people or groups, ask “what is this person likely to do next?”, then give a prediction plus the visual reason that supports it.

Verify the current format in the official CELPIP Speaking Pro: Target 9+ 2026 study pack and the CELPIP-General test format.

CELPIP Speaking Task 4 format

• Task: Predict what will happen next in an illustrated scene
• Preparation time: 30 seconds
• Speaking time: 60 seconds

Task 4 is not a second Task 3. You may briefly identify the person or action you are referring to, but your main job is prediction.

The strongest prediction pattern

A practical sentence pattern is:

visible evidence → prediction → likely result

Example:

“The little boy is reaching toward the ice cream counter, so I think he is going to ask his father to buy him a cone. His father will probably agree because they are already waiting in line.”

The prediction is easy to follow because it is anchored in what is visible.

What CELPIP raters evaluate

Content and coherence

Give several distinct predictions and develop them enough that the listener understands why they are plausible. Avoid repeating the same prediction with different wording.

Vocabulary

Use action-specific language and varied expressions of probability when they fit naturally:

• will probably;
• is likely to;
• might;
• may;
• looks as though;
• seems ready to;
• is probably going to.

You do not need to force every future form into one response.

Listenability

Keep each prediction relatively compact. Long conditional sentences can become difficult to control under a 60-second timer. Clear grammar and steady delivery matter more than complexity for its own sake.

Task fulfillment

Predictions should be connected to the scene. Completely unrelated future events waste time and weaken relevance.

A 30-second planning method

Choose three prediction targets:

Person A → current action → next action
Person B → current action → next action
Group C → current action → next action

If time remains, add one likely feeling or consequence.

For example:

runner → tied shoe → rejoin race
child → dropped balloon → cry / parent help
vendor → packing boxes → close stall

A flexible Task 4 template

Opening

• “Based on what is happening in the picture, I think several things are likely to happen next.”
• “There are a few clear clues about what these people will probably do next.”

Prediction 1

Identify the person briefly, then predict.

• “The woman near the entrance is holding an umbrella and looking at the sky, so she will probably open it in a moment.”

Prediction 2

• “Meanwhile…”
• “On the other side…”
• “Another likely event is…”

Prediction 3

Add another visually supported outcome.

Closing

A closing is optional if your final prediction is complete. Do not sacrifice a useful prediction just to recite a memorized ending.

Original practice scene

Imagine a farmers' market near closing time. A vendor is stacking empty fruit boxes. A customer is reaching into a wallet beside a basket of apples. A child is holding a balloon whose string is slipping through her hand. In the background, dark clouds are moving over the market and two shoppers are already putting on jackets.

Original sample response

A few things are likely to happen next in this market. First, the customer beside the apple stand is reaching into his wallet, so he will probably pay for the fruit and leave with the basket. The vendor is already stacking empty boxes, which suggests that she is going to close her stall once the last customers finish shopping. The little girl's balloon string is slipping out of her hand, so the balloon may float away and she might become upset unless an adult catches it quickly. Finally, the sky is getting very dark and some shoppers are putting on jackets, so I think it will start raining soon and people will probably leave the market in a hurry.

Why this response works

• Every prediction has a visible clue.
• The answer covers different people instead of repeating one event.
• Probability language varies naturally.
• The sequence is easy to follow.
• Predictions remain realistic.

Common Task 4 mistakes

Describing instead of predicting

Do not spend most of the minute saying what people are currently doing. Use current actions only as evidence.

Making dramatic unsupported predictions

A person holding a coffee is unlikely to “move to another country next week.” Stay close to the scene.

Repeating “will” in every sentence

Mix future expressions only when natural. Accuracy matters more than variety for its own sake.

Predicting the same event for everyone

Choose different people or groups and produce separate outcomes.

Ignoring visual evidence

A prediction becomes stronger when the listener can understand why you made it.

Level 9-focused self-check

Content/Coherence — Did I give several clear, connected predictions?
Vocabulary — Did I describe people/actions precisely and vary probability language naturally?
Listenability — Was the response clear at a controlled pace?
Task Fulfillment — Were the predictions realistic and connected to the illustration?

Practice Task 4

Use CELPIP Speaking Practice for timed prediction prompts. Review all eight tasks on the 2026 CELPIP Speaking Guide, and test your pacing in a full CELPIP mock test.

Frequently asked questions

How much preparation time is there for Task 4?

The current official format gives you 30 seconds.

How long do I speak?

You have 60 seconds.

Is Task 4 the same as describing the picture?

No. Task 3 describes what is happening now; Task 4 focuses on what you think will happen next.

Do predictions need to be certain?

No. Use language such as “probably,” “might,” or “is likely to” when the outcome is uncertain.

How many predictions should I give?

There is no magic required number. Aim to cover several distinct, well-supported predictions rather than rushing through many weak ones.