For CELPIP Speaking Task 4, explain what is likely to happen next in a scene and connect each prediction to a visible clue. A useful response goes beyond naming a future event: it identifies the person or object involved, explains the reason for the prediction, and develops a plausible immediate consequence.
The supplied official Listening and Speaking Study Guide describes Task 4 as Making Predictions, using the same illustration as the preceding scene-description task. It provides thirty seconds of preparation and sixty seconds of recording time. Follow the actual prompt and timer. The official test format confirms the task's place in the Speaking component.
All examples here are original teaching material. The scene drills are written descriptions, not actual pictures or authentic exam questions. They help you practise prediction language and reasoning; combine them with real image practice so that you also learn to notice visual clues independently.
Shift your attention from the present to the next event
Task 3 and Task 4 require different communication. Describing a scene answers “What is happening?” Predicting answers “What is likely to happen next?” If you repeat a detailed description without moving forward in time, the response may be clear English but incomplete for the prediction task.
Consider a person carrying several boxes toward a closed door. A description says, “A man is carrying boxes near the entrance.” A prediction says, “He will probably have to put the boxes down to open the door, unless someone nearby holds it for him.” The second statement uses the current situation to anticipate a next step.
Your prediction does not have to be surprising. Ordinary consequences are often easier to justify. A customer approaching a counter may place an order. A volunteer holding a sign beside a queue may direct people to another entrance. The quality of the explanation matters more than how dramatic the imagined event is.
Avoid treating the scene as a puzzle with one hidden future. Different predictions can be plausible, and the supplied guide emphasizes communicating predictions effectively rather than finding a single correct answer. Your task is to develop a reasonable account in clear language, using the scene as the starting point.
A useful preparation question is “What is already in motion?” Look for approaching people, unstable objects, unfinished tasks, reactions, or visible preparations. These details naturally suggest a near future and make your explanation more specific than a general statement that everyone will eventually go home.
Build a clue-to-consequence chain
A developed prediction contains a clue, an expected event, and sometimes a consequence. The clue grounds your idea. The expected event answers the task. The consequence extends it enough to show coherent reasoning. You can present these elements in different orders as long as their relationship is clear.
For example: “The cyclist is looking toward the shop window instead of the path, so she might not notice the person stepping out ahead of her. I think she'll brake suddenly, and the pedestrian will probably stop to let her pass.” This prediction remains near the visible action and explains why the speaker expects it.
Compare “The cyclist will have an accident” with the developed version. The shorter statement may be possible, but it lacks support and jumps to a severe result. The developed answer explains attention, movement, and a likely response. It also avoids treating an uncertain outcome as inevitable.
You do not need three sentences for every prediction. Sometimes a single sentence can include the whole chain: “Because the volunteer is already reaching for the fallen papers, she'll probably collect them before the wind carries them farther away.” The amount of development should fit the complexity of the action.
When reviewing your answer, underline the evidence behind each prediction. If you cannot identify a visible clue, ask whether the idea belongs in the response. A prediction about someone receiving a promotion next month may be imaginative, but it is difficult to justify from an ordinary scene in a workplace.
Choose a useful time horizon
The phrase “what will happen next” encourages attention to the near future. Immediate developments usually connect most directly to visible actions. Someone reaching for a cup may pick it up. Someone standing on an unstable stool may regain balance or ask for help. A queue moving toward an open desk may begin being served.
Long-term predictions often require more invented information. A child looking at a telescope might become interested in astronomy, but predicting a future scientific career goes far beyond the scene. You can mention a modest consequence if it follows naturally, yet the main response should remain anchored in what is likely to unfold shortly.
The appropriate horizon depends on the scene. In a picture of people preparing a room, the next meaningful event might be the arrival of guests. In a picture of a ball flying toward a window, the next event is much more immediate. Use the action's natural pace rather than forcing every prediction into the same number of seconds.
Do not extend a chain simply to fill time. “The cup will fall, the floor will get wet, someone will slip, an ambulance will arrive, and the shop will close” contains several unsupported leaps. A stronger response might predict the spill and explain who is likely to clean it up.
Practice by drawing a boundary around your prediction. Ask what the scene clearly prepares and where your imagination begins adding unrelated events. Keep the useful continuation and remove distant consequences that depend on assumptions the listener cannot evaluate.
Original worked prediction: a community event setup
This original text drill describes a scene for language practice. Several volunteers are preparing an outdoor information booth. A stack of leaflets is sliding toward the edge of a table. One volunteer is reaching toward it. Another is carrying a folded sign toward the entrance. Two visitors are approaching the booth, and a third volunteer is opening a box of maps. Dark clouds are visible in the distance, but the area is currently dry.
An original response might be:
I think the volunteer beside the table will catch most of the leaflets before they fall, because she is already reaching toward the stack. She'll probably place something heavier on top afterward to stop them from moving again.
The person carrying the sign is likely to set it near the entrance so that visitors can find the booth more easily. Once the two people approaching the table arrive, the volunteer opening the box will probably offer them a map and explain where the activities are taking place.
There may also be a change in the weather. Since dark clouds are approaching, the group might move the paper materials under cover or keep them inside the box. I wouldn't assume the event will be cancelled, but they will probably want to protect the information they have prepared.
This response develops three different areas: a small immediate problem, the event's intended activity, and a possible weather adjustment. It does not describe every object before making a prediction. Present details appear only where they support a likely future action.
The weather point also demonstrates restraint. Dark clouds support the possibility of rain, but they do not establish that the whole event will be cancelled. The speaker predicts a practical precaution and explicitly avoids a much larger unsupported conclusion. This makes the reasoning easier to accept.
Use future forms to express different meanings
“Will probably” is useful for an expected event: “The visitor will probably ask for directions.” “Is going to” can express a prediction based on a visible developing situation: “That stack is going to fall if nobody catches it.” “Might” and “may” mark a less certain possibility. These forms should reflect your intended meaning rather than appear as a checklist.
“Is about to” suggests an event is very close to happening. Use it when the visible action supports that timing, such as a person with a hand already on a door handle. Do not use it for a distant or vague possibility. “She is about to change careers” is unlikely to be supported by a simple scene.
“Is likely to” provides another way to express probability: “The organizer is likely to direct the late arrivals to the empty seats.” You can also use “I expect” or “I think” when those phrases sound natural. Avoid beginning every sentence with the same expression, but do not force unnecessary variation that makes your speech hesitant.
Conditional language helps explain a possible branch: “If the wind gets stronger, they may move the display indoors.” The condition makes your uncertainty explicit. Keep the main clause grammatically clear. A reliable pattern is “If + present form, will or may + base verb” when discussing a real future possibility.
Avoid a sentence such as “If the papers will fall, the volunteer will catches them.” Repair it to “If the papers fall, the volunteer will try to catch them.” The correction clarifies the relationship and removes agreement problems. Practise the whole pattern with several ordinary actions so that you can retrieve it naturally.
Calibrate confidence to the evidence
Confidence language should match how much the scene supports the event. A person holding a key at a locked door is likely to unlock it. A person looking at a notice might be reading, deciding what to do, or searching for a name. The second case deserves more cautious wording than the first.
You can be specific without claiming certainty. “The man may ask the volunteer for help because he appears to be studying the map without moving” gives a plausible event and a reason. “Maybe something will happen with the man” is cautious but uninformative. Uncertainty should qualify a useful prediction, not replace one.
Likewise, certainty alone does not make an answer strong. “The manager will definitely fire that worker” can sound decisive while relying entirely on invented relationships and consequences. Replace it with what the scene supports: “The person pointing at the damaged display will probably ask for help repairing it.”
Use one clear probability expression rather than stacking several. “He might possibly perhaps be going to ask for assistance” becomes difficult to process. “He may ask for assistance” communicates the same uncertainty efficiently. This leaves more time to explain why you expect the event.
In practice, sort your ideas into strong expectation, reasonable possibility, and unsupported guess. Use the first two categories for your response and discard the third. The categories are a learning tool, not an official scoring system. They train you to connect language choice with evidence quality.
Select predictions that add different information
Three predictions about the same falling object may become repetitive. If the scene provides several activities, choose predictions that show different developments. One might resolve a small problem, another might continue an intended task, and a third might describe someone's response to an approaching event.
Suppose a market scene includes a customer holding coins, a vendor lifting a bag of produce, and a child reaching toward an apple. You could predict the purchase and the handover as one connected event. Treating each tiny movement as a separate major point may make the response feel artificially divided.
Look for a second area, such as a delivery person approaching with a cart. That could support a prediction about unloading supplies or making room near the stall. The new point expands the listener's understanding of the scene rather than restating the same transaction.
There is no required number of predictions in this article's method. Two well-developed chains can be more informative than five unsupported statements. The task's recording limit means you need to balance breadth and explanation. Choose the number that lets you identify people, support expectations, and finish coherently.
If one prediction depends on another, explain the dependency. “After the customer pays, the vendor will probably hand over the bag.” If the predictions are independent, use a simple location transition: “On the other side of the scene…” The transition should tell the listener whether you are continuing an event or changing focus.
Keep people and objects identifiable
Predictions can become confusing when several people are described only as “he,” “she,” or “they.” Begin each new chain with an identifying phrase: “The person carrying the ladder,” “the child near the fountain,” or “the volunteer beside the table.” Once the reference is established, a pronoun can help the sentence flow.
Avoid guessing names or relationships unless the prompt provides them. The person beside a child might be a parent, teacher, or another visitor. You can still predict a helpful action: “The adult standing beside the child may pick up the dropped bag.” Accurate identification is more important than assigning a social role.
When two objects are involved, name them clearly before using “it.” “The cup is near the edge of the tray, so it may fall” could refer to either the cup or the tray. “The cup may slide off the tray” removes the ambiguity and describes the movement more precisely.
You can use spatial references from your earlier scene description, but do not assume every detail remains clear to the listener. A short reminder is enough: “At the rear doorway, the two waiting visitors will probably enter next.” This re-establishes the location while moving directly to the prediction.
During transcript review, circle every pronoun and draw a line to the noun it refers to. If you cannot connect it confidently, replace the pronoun or reorganize the sentence. This simple exercise often improves clarity more than adding more elaborate future expressions.
Repair predictions that become stories
An overdeveloped response may turn one clue into an entire fictional narrative. A person holding a suitcase becomes a traveller who misses a flight, loses a job, moves to another city, and changes careers. Each step moves farther from the scene and consumes time without improving the immediate prediction.
To repair this, return to the last supported event. If the person is approaching a taxi, predict that they will load the suitcase and leave. If the driver is looking at a watch, you might add that the traveller will hurry. Stop before the chain requires information about destinations, employment, or personal history that the image does not provide.
Another problem is repeating the scene for too long before predicting. “There is a person with a box, and the box is large, and the person is close to a table” can become “The person carrying the large box will probably place it on the nearby table.” The revised sentence includes useful description while answering the actual question.
Sometimes a prediction is plausible but unexplained. “They will move indoors” needs a reason. Add the visible cue: “Because the wind is blowing the papers across the table, they may move the information desk indoors.” The change is small, but the listener can now follow your reasoning.
A final repair concerns exaggerated certainty. Replace “Everyone will panic” with a more measured action when appropriate: “Some visitors may step back, while the organizer asks people to make room.” Specific behaviour is easier to visualize and usually more useful than a sweeping emotional claim.
Practise with paired observation and prediction
Use an actual image from an appropriate practice source or a picture you have permission to view. First, make a short description of the present scene. Then create a separate prediction response. Keeping the recordings separate helps you notice when you accidentally substitute description for prediction.
After the prediction recording, list each future event you mentioned and the clue supporting it. You should be able to make a two-column record: visible evidence and predicted development. If a row has no evidence, revise or remove that idea. If several rows use the same clue, check whether the response is becoming repetitive.
For a language drill, change the confidence level while preserving the event. Say “The visitor will probably ask for directions,” then “The visitor may ask for directions.” Explain what additional visual detail would justify the stronger version. This connects grammar practice to meaning instead of treating future forms as interchangeable decorations.
For a reasoning drill, write two plausible futures from one clue. A person reaching toward falling papers might catch them, or some papers might land before the person reaches them. Develop either possibility without insisting that one is the only acceptable future. Then choose the version you can explain most clearly in a timed response.
Finally, practise with a new image. Repeating one response can improve delivery, but transfer requires fresh observation. Use the same evidence-to-consequence method while allowing the number, order, and language of predictions to change with the scene.
Review meaning and delivery separately
The official results information identifies Content/Coherence, Vocabulary, Listenability, and Task Fulfillment as Speaking dimensions. For Task 4, a useful content review asks whether the predictions are developed and connected. A task review asks whether you actually discussed likely next events rather than only the current picture.
Vocabulary review should focus on accurate actions and degrees of certainty. “The volunteer will secure the sign” is useful only if you can say it naturally and mean fastening it safely. A simpler “tie the sign to the fence” may be more precise in context. Choose language that communicates the visible action rather than language that merely sounds advanced.
For listenability, check whether each prediction forms a complete thought before you move to another area. Pause briefly between chains so the listener can reset attention. Avoid rushing through several consequences with no sentence boundary. Clear grouping makes even a detailed response easier to follow.
Record a second attempt with one change in mind, such as adding evidence after every prediction or reducing unsupported consequences. Compare that feature directly. Self-review can guide practice, but it does not establish an official level or guarantee a future test result.
Frequently asked questions
Is there one correct prediction for each person?
No single future needs to be discovered. The supplied guide treats the task as an opportunity to communicate predictions effectively. Choose plausible developments you can explain clearly. Visible clues help your answer remain coherent and prevent it from drifting into an unrelated story.
Can I use “might” instead of “will”?
Yes, when you mean the event is possible rather than strongly expected. Use probability language deliberately. A response can include several degrees of confidence, provided each matches the evidence and the listener understands what you expect to happen.
How much description should I include?
Include enough to identify the person, object, and clue behind a prediction. You do not need to repeat your full Task 3 description. A phrase such as “the volunteer reaching for the papers” can establish the relevant detail while keeping the focus on the future.
Can written drills replace picture practice?
No. Written drills support grammar and reasoning, but they provide selected clues in words. Actual image practice is necessary to develop observation, selection, and spatial reference. Use both types of activity for different learning purposes.
Sources and further reading
- CELPIP, Listening and Speaking Study Guide, seventh edition, December 2025: Speaking overview and Task 4, Making Predictions. Supplied source material.
- Official CELPIP test format and scoring
- Official CELPIP results and performance dimensions
- Official CELPIP free preparation resources