CELPIP Speaking Task 4: Making Predictions

CELPIP Speaking Task 4 format and timing

Move from describing current actions to explaining likely next events. A strong practice response makes each prediction traceable to something in the picture. For example, a leaning stack supports a prediction that it might fall; it does not support a story about why someone bought the pots.

The real test reuses the Task 3 image. Our exercise also reuses its scene, but supplies it in words because there is no image on this page. This supports language practice while leaving out the visual interpretation you would do on the test. Keep that difference in mind when reviewing your performance.

The timings above are test timings. The response plan below is our practice suggestion; you can adjust its divisions to suit your delivery.

How Task 4 uses the four speaking dimensions

• Content/Coherence: connect a visible clue to a likely consequence. Develop a few predictions instead of listing disconnected guesses.
• Vocabulary: express probability with language such as “likely”, “might”, and “probably”. Choose action verbs that make the predicted event specific.
• Listenability: control future and conditional structures, and pause between predictions so the listener can follow your reasoning.
• Task Fulfillment: explain what may happen next in this scene. Repeating the Task 3 description does not complete the prediction task.

Different predictions can be reasonable when they fit the same evidence. Your job is to make the connection clear rather than invent the most dramatic outcome.

A response structure for making predictions

1. Preparation, 30 seconds: pair visible clues with possible next actions. Choose the clearest connections and keep the time horizon close to the scene.
2. Opening, about 10 seconds: introduce your most obvious prediction and its clue.
3. Development, about 40 seconds: add further predictions with reasons. Use a conditional sentence if an outcome depends on someone acting.
4. Ending, about 10 seconds: describe a likely overall change, such as the group finishing work or moving to shelter.

A useful rehearsal note is “clue → event → reason”. You can say those elements in a natural order; the listener does not need to hear the planning labels.

Original Task 4 practice prompt

Imagine the picture shows a community garden in Winnipeg. In the foreground, a volunteer in a green jacket holds a watering can beside newly planted seedlings. To the left, a child reaches towards a leaning stack of empty pots while an adult turns towards the child. At the back, two neighbours carry a wooden bench towards a clear patch beside the fence. A wheelbarrow blocks part of the central path. Dark clouds are gathering overhead. This is the same scene used in Task 3. Explain what will probably happen next and give reasons based on details in the scene.

The real task uses the same image as Task 3; this exercise uses the same text substitute. Consider the leaning pots, the turning adult, the bench, and the clouds. Choose plausible immediate developments and explain which clue led you to each prediction.

Task 4 sample answer

Why this sample works

• Content/Coherence: the response links each prediction to a scene detail, including the leaning pots and the clear space by the fence.
• Vocabulary: “steady”, “route”, and “shelter” describe likely actions and needs precisely. Probability language varies without making the answer sound uncertain about everything.
• Listenability: conditional sentences explain the effect of rain. Each paragraph gives the speaker a natural boundary between ideas.
• Task Fulfillment: the focus remains on likely next events rather than repeating a static description.

The weather prediction allows more than one outcome, which is reasonable because dark clouds alone do not establish exactly when rain will start.

Common CELPIP Speaking Task 4 mistakes

• Retelling everything visible before reaching the first prediction.
• Predicting distant events, such as next year’s harvest, without support from the scene.
• Using certainty for every outcome even when the clue allows several possibilities.
• Giving a prediction without naming the detail that makes it plausible.
• Inventing an accident or argument merely to make the response more exciting.
• Using complicated future forms that you cannot finish clearly under time pressure.

After recording, write down the evidence for each prediction. If you cannot identify any, replace that prediction with one grounded in a visible action or object.

How to review your timed response

Try the prompt before reading the sample, then listen to your own recording without following a transcript. Mark the point where your meaning becomes difficult to follow. Was the problem an unfinished idea, an unclear reference, a long pause, or a sentence that became too complicated? Choose one issue to work on in your next attempt.

Then check the transcript for relevance and useful detail. A readable transcript cannot establish how your pronunciation sounded. CLB Mock, the app CELPIP Practice Test: AI Score by Certloom Learning, gives an estimated CELPIP level (≈CLB) range from a transcript and timing metrics. Listenability, including pronunciation, is only estimated from fluency metrics. Use that feedback alongside listening to your recording.

Choose your next speaking exercise

Revisit Task 3’s scene description to compare present observations with predictions. Then practise supporting a choice in Task 5: comparing and persuading.

Want more examples first? See Task 4 questions with sample answers, or plan with the speaking templates.

Use the speaking overview to compare all eight tasks, and the score chart to understand the level scale. Read how AI scoring estimates work before interpreting app feedback. This is an unofficial practice guide with original exercises and samples. Its sample band is our teaching estimate, not a test result. Keep the useful structure, then change the details and wording so your next response is your own.