CELPIP General
CELPIP Speaking Task 8: Describing an Unusual Situation | Complete Guide
CELPIP Speaking Task 8 at a Glance Aspect Detail Task Describing an Unusual Situation Prep time 30 seconds …
CELPIP Speaking Task 4 at a Glance
| Aspect | Detail |
|---|---|
| Task Name | Making Predictions |
| Position | Task 4 of 8 in the CELPIP Speaking section |
| Prep Time | 30 seconds |
| Speaking Time | 60 seconds |
| Format | Same image as Task 3, predict what happens next |
| Key Function | Predicting, speculating, using conditional language |
| Connected to | Task 3 (same image, current description) |
| Scoring | Content/Coherence, Vocabulary, Listenability, Task Fulfilment |
Tasks 4 and 5 step up the complexity of CELPIP Speaking. Instead of just describing or advising, you now need to predict what might happen and reason from visible clues. These tasks test your ability to think hypothetically and argue a position. They are the skills that separate CELPIP 7 from CELPIP 9.
Task 4 is the easier of the two, but only if you walk in with the right language pattern locked in. Students who improvise future tense often fall back on “I think” five times in 60 seconds and score below their actual English level.
What Is CELPIP Speaking Task 4?
You see the same image from Task 3 on screen. Your job is to predict what will happen next, or what might happen in the future. You need to use conditional language and logical reasoning.
You are not describing what IS happening. You are predicting what WILL or MIGHT happen.
Tip
The language you need here is fundamentally different from Task 3. You are not saying what is happening. You are saying what will or might happen. Phrases like ‘I think they will probably…’, ‘It is likely that…’, ‘Based on what I see, I predict…’ should come out automatically.
Typical Task 4 Image Triggers
The image always contains visible clues that point toward predictable next events:
- A family packing boxes (moving day predictions)
- A couple at an airport check-in (travel predictions)
- A young person reading a thick textbook (exam or study predictions)
- A worker carrying tools toward a building (repair or construction predictions)
- A child opening a wrapped gift (reaction predictions)
- Dark clouds gathering above a picnic (weather change predictions)
Each scene gives you 2 to 3 clear prediction paths. Pick the strongest two.
The Exact Format with a Worked Sample Prompt
Sample CELPIP Task 4 Prompt
The image you described in Task 3 stays on the screen. What do you think will happen next? Make predictions about what these people will do, and why you think so.
(Image: A family of four packing cardboard boxes in a half-empty living room. The parents look tired. The children are putting toys into a small box. A ‘For Sale’ sign is visible through the window.)
You see the image. You hear and read the instruction. A 30-second prep timer counts down. Then you record for 60 seconds.
How CELPIP Speaking Is Scored
Every speaking task is scored on the same four dimensions.
| Dimension | What It Means | What Scores Well in Task 4 |
|---|---|---|
| Content/Coherence | Did you address the task fully and logically? | 2 to 3 distinct predictions, each with a reason |
| Vocabulary | Range and accuracy of words used | Variety in future-tense language, no repetition of “I think” |
| Listenability | How easy is it to follow your speech? | Confident future forms, natural conditional language |
| Task Fulfilment | Did you do what the task asked? | Predicted (did not describe), tied predictions to visible clues |
Info
CELPIP uses human raters for Speaking, not AI. This means they can recognise creative predictions and reward well-reasoned speculation. They also flag responses that drift back into description (Task 3 mode) and penalise them on Task Fulfilment.
Scoring Bands for Task 4
| Dimension | Score 9-12 (CLB 9+) | Score 7-8 (CLB 7-8) | Score 5-6 (CLB 5-6) |
|---|---|---|---|
| Content/Coherence | 3 distinct predictions, each tied to a visible clue and a reason. | 2 predictions, mostly tied to clues. Reasons may be light. | 1 prediction repeated, or predictions with no reasoning. |
| Vocabulary | Wide range of future forms: will, going to, might, could, is likely to, expects. | Some variety, but “I think” or “will” used heavily. | Limited future-tense forms, mostly “I think” or “they will”. |
| Listenability | Confident future and conditional forms, clear contractions. | Generally clear, occasional grammar slips. | Frequent grammar errors with future tense. |
| Task Fulfilment | Pure prediction tied to visible evidence. Mentions reasoning. | Mostly prediction, some drift into description. | Mixes description and prediction. Sometimes only describes. |
Step-by-Step Approach: 30-Second Prep + 60-Second Speak
The 30-Second Prep Window
The image is already familiar from Task 3. Your prep is about reasoning, not observation.
- Pick 3 visible clues from the image (10 sec). Boxes, expressions, items in hand, time of day.
- For each clue, generate one prediction (15 sec). What does this clue tell you about what happens next?
- Pick your 2 strongest predictions (5 sec). Set the third aside as backup.
Warning
Do not re-describe the image. That is Task 3 work. Task 4 prep is about converting observations into “they will probably…” statements. Spend zero seconds on description planning.
The 60-Second Speaking Window: Marvel’s Three-Beat Structure
| Section | Timing | What to Cover |
|---|---|---|
| Brief setup | 10 sec | One sentence to anchor the prediction in the image |
| Predictions 2 to 3 | 35 to 40 sec | Two strong predictions with visible-clue reasoning |
| Closing reasoning | 10 to 15 sec | Explain why you say all this, give a broader insight |
You should be making your first prediction by the 10-second mark. Anything later means you over-described.
Verbatim CLB 10 Sample Response (with Annotations)
Here is a model response to the family-packing-boxes sample prompt.
Model Response: Family Packing Boxes
(Brief setup) In this image, I can see a family packing boxes in a half-empty living room. They seem to be preparing to move, especially given the ‘For Sale’ sign outside the window.
(Prediction 1 + reasoning) I think they will probably move to a new city for a job opportunity. The fact that the parents look tired but determined suggests they have been planning this for a while. They might feel excited but also nervous about starting over in an unfamiliar place.
(Prediction 2 + reasoning) I also predict the children will initially struggle to make new friends and might miss their current school, but they will likely adapt within a few months. Children at that age usually settle quickly once they find a routine and a couple of friendly classmates.
(Closing reasoning) I say all this because moving is always challenging at first, but most families find their rhythm after the first six months. There is a good chance that within a year, this house move will feel like one of the best decisions they made.
Why This Response Scores High
- Content/Coherence: Two distinct predictions (about the parents’ situation, about the children’s adjustment) plus broader closing reasoning. Each prediction is tied to a visible clue.
- Vocabulary: Wide future-tense range: “will probably”, “might feel”, “will likely”, “there is a good chance”. No repetition of “I think”.
- Listenability: Confident future forms throughout. Natural conditional language. Pauses fall at sentence breaks.
- Task Fulfilment: Almost zero re-description. Goes straight into predictions tied to visible evidence (the boxes, the For Sale sign, the parents’ faces).
Sentence Stems and Opening Phrases
Marvel’s recommended phrase bank for Task 4. Pick three or four per response.
Brief Setup (10 sec)
- “In this image, I can see…”
- “Based on what I described earlier…”
- “Looking at this scene, it seems clear that…”
- “Given the visible clues in the picture…”
Introducing Predictions (Future-Tense Variety)
- “I think they will probably…”
- “It is likely that…”
- “They might…”
- “Based on what I see, I predict…”
- “There is a good chance that…”
- “I expect that…”
- “Chances are they will…”
- “It would not surprise me if…”
Tying Predictions to Visible Clues
- “Given that…”
- “Because they look…”
- “The fact that… suggests…”
- “Judging by the…”
- “Considering they are…”
Closing Reasoning
- “I say this because…”
- “All of this points to…”
- “Overall, I would say…”
- “In the end, my best guess is…”
Tip
Use four different future-tense forms in one Task 4 response. “Will”, “going to”, “might”, and “is likely to” in a single 60-second window. This is the single most reliable way to push your Vocabulary score from CLB 8 to CLB 9. Variety in speculative language impresses raters.
Common Mistakes That Cost Marks
| Mistake | What It Sounds Like | Score Impact |
|---|---|---|
| Re-describing the image | “In the foreground there is a woman, behind her there are…” (Task 3 mode) | Task Fulfilment drops sharply |
| Repeating “I think will” | “I think they will… I think they will… I think they will…” | Vocabulary drops |
| Predictions with no reasoning | “They will move. They will be sad. They will adapt.” | Content/Coherence drops |
| One prediction stretched | Talking about one move for 60 seconds with no second prediction | Content/Coherence drops |
| Wrong tense (present) | “They are moving and they have new friends and they like it” | Listenability and Task Fulfilment drop |
| Generic predictions | “Maybe they will be happy. Maybe they will be sad.” | Vocabulary and Content drop |
| Stopping at 35 seconds | Trailing off after one prediction | Content/Coherence drops |
| Memorised template | “In the future I predict three things will happen, firstly…” | Task Fulfilment drops |
Warning
The biggest Task 4 mistake: treating it as Task 3 part two. You are not describing what is happening. You are predicting what will happen. If you find yourself saying “is walking” or “are sitting”, stop and reframe to “will probably walk” or “are going to sit”.
CELPIP vs IELTS GT vs PTE Core Speaking
| Aspect | CELPIP General | IELTS General Training | PTE Core Speaking |
|---|---|---|---|
| Format | Recorded into microphone, 8 separate tasks | Face-to-face with examiner, 3 parts | Recorded into microphone |
| Prediction task | Task 4: explicit prediction from image | Sometimes Part 3 (“Do you think this will continue?”) | Not directly tested |
| Prep time | 30 sec for Task 4 | None for IELTS speaking | None or very short |
| Speak time | 60 sec for Task 4 | Variable | Variable |
| Examiner | None (recording only) | Yes, live examiner | None (AI scored) |
| Visual-based prediction | Yes (same image as Task 3) | No | No |
| Scoring criteria | Content, Vocabulary, Listenability, Task Fulfilment | Fluency, Lexical Resource, Grammar, Pronunciation | AI multi-trait |
| Rater type | Trained human rater | Trained human examiner | Automated AI |
Info
Key takeaway: Task 4 has no direct equivalent in either IELTS GT or PTE Core. The closest IELTS analogue is a Part 3 prediction question, but those are conversational and prompted by the examiner. PTE has no equivalent at all. This means many candidates arrive at Task 4 underprepared for sustained 60-second prediction speech, even if they have strong English. Targeted practice matters.
How to Practise CELPIP Task 4 Properly
What Effective Practice Looks Like
- A bank of unfamiliar image prompts (so you cannot pre-rehearse predictions)
- A 30-second prep timer
- A microphone that locks at 60 seconds
- Replay with image alongside (so you can review which clues you used)
- Feedback specifically on future-tense variety and conditional language
- Paired Task 3 and Task 4 practice on the same image
A Weekly Practice Plan
| Day | Activity | Time |
|---|---|---|
| Mon | 4 fresh Task 4 prompts, recorded, no notes | 12 min |
| Tue | Listen back, count repetitions of “I think” and “will” | 15 min |
| Wed | Repeat Monday’s prompts using 4 different future forms each | 12 min |
| Thu | Paired Task 3 + Task 4 on three new images, back to back | 18 min |
| Fri | Mix Tasks 1, 2, 3, 4 to build switching agility | 25 min |
| Sat | Future-tense drill: 30 sentences using 6 different forms | 10 min |
| Sun | Full Speaking section practice (all 8 tasks) | 25 min |
Three More Sample Scenes with Prediction Plans
Sample Scene 2: Airport Check-in
Image: A young couple at an airport check-in counter, surrounded by luggage, both holding passports, looking nervous.
Prep plan (30 sec):
- Clue 1: amount of luggage suggests a long trip
- Clue 2: their nervous faces suggest first major trip or anxious flyers
- Prediction 1: they are going on a long holiday or relocating abroad
- Prediction 2: once they board, they will feel relieved and excited
Opening 30 seconds:
Based on this image, I think this couple is about to begin a fairly long trip, probably a vacation or maybe even a move abroad given the amount of luggage they are carrying. The nervous expressions on their faces suggest they might not fly very often, so the next hour will probably feel stressful. Once they get through security though, I expect they will start to relax.
Sample Scene 3: Student Studying Late
Image: A university student at a desk surrounded by textbooks, an empty coffee mug, late-night lamp light, the clock reads 11:30 pm.
Prep plan (30 sec):
- Clue 1: stacked textbooks suggest exam tomorrow
- Clue 2: late hour and empty mug suggest long study session ongoing
- Prediction 1: he will likely finish reviewing in the next hour
- Prediction 2: he might struggle on the exam tomorrow due to lack of sleep
Sample Scene 4: Rain Approaching a Picnic
Image: A family picnicking in a park, kites flying, but dark clouds gathering in the distance.
Prep plan (30 sec):
- Clue 1: dark clouds suggest imminent rain
- Clue 2: family unaware, kids playing freely
- Prediction 1: the family will probably need to pack up within 15 to 20 minutes
- Prediction 2: the children might be disappointed when they have to stop playing
Vocabulary Bank for Task 4
| Function | Form | Example |
|---|---|---|
| High certainty | will definitely | “They will definitely arrive on time.” |
| Moderate certainty | will probably | “They will probably move next month.” |
| Moderate uncertainty | might | “They might decide to wait.” |
| Lower certainty | could | “It could rain before evening.” |
| Likelihood phrase | is likely to | “She is likely to accept the job.” |
| Probability phrase | there is a good chance that | “There is a good chance that he will pass.” |
| Conditional | if + present, will + verb | “If they pack quickly, they will avoid the rain.” |
| Hedging | I expect that | “I expect that they will settle in quickly.” |
Tip
Aim to use at least four of these in every Task 4 response. Each one signals a different degree of certainty, which makes your speech sound nuanced and natural. Raters specifically listen for this.
Recovering from a Bad Start in Task 4
| Situation | What to Do |
|---|---|
| You blank at start | Anchor with “Based on this image, I would predict that…” then begin |
| You drift back to describing | Self-correct: “Actually, what I think will happen next is…” |
| You repeat “I think” | Switch to “It is likely that…”, “There is a chance that…” |
| You finish at 40 seconds | Add a closing reasoning line: “All of this points to a transition period for them.” |
| You feel rushed at 55 sec | Cut to a one-line big-picture prediction: “Overall, things will probably work out for them.” |
Connecting Task 3 and Task 4 as One Exercise
Marvel Education’s strongest test takers treat Tasks 3 and 4 as a single two-minute exercise on one image. Here is the connected playbook:
- During Task 3 prep, tag 2 to 3 elements as “Task 4 fuel” (boxes, expressions, weather, unfinished actions).
- In Task 3, describe everything but explicitly mention those tagged elements with detail.
- When Task 3 ends, your prediction material is already loaded in your head.
- In Task 4 prep, you only need to convert those tagged observations into future-tense predictions.
- In Task 4, refer back: “I mentioned the For Sale sign earlier, and based on that, I predict…”
This cross-task strategy is the difference between an improvised Task 4 and a planned one. It often raises overall Speaking scores by half a CLB level.
Practise on the Real CELPIP Interface
Marvel Education’s CELPIP practice software replicates the real test interface:
- A bank of 200+ realistic scene images with paired Task 3 and Task 4 prompts
- 30-second prep timer matching the official one
- Microphone capture auto-stopping at 60 seconds
- Replay alongside the original image
- AI feedback on future-tense variety, reasoning quality, and prediction structure
- CLB 10 model responses written by Marvel Education’s CELPIP teachers
- Cross-task analytics that show how your Task 3 and Task 4 responses link
Start Free Trial of Marvel CELPIP Practice Software
Quick Recap: The Task 4 Playbook
- The same image stays on screen from Task 3. Do not re-study it.
- Spend 30-sec prep converting observations into predictions, not describing.
- Open with a 10-second anchor sentence, then jump straight into predictions.
- Make two strong predictions, each tied to a visible clue.
- Use four different future-tense forms: will, going to, might, is likely to.
- Avoid repeating “I think” more than twice in 60 seconds.
- Tie every prediction to evidence: “Given that…”, “Because they look…”.
- Avoid describing. The moment you say “is walking”, you are in the wrong task.
- Close with broader reasoning: “All of this points to…”, “Overall, my best guess is…”.
- Aim to finish with a clear, confident closing line just before 60 seconds.
Frequently Asked Questions
You get 30 seconds to prepare and 60 seconds to speak. The image you saw in Task 3 stays on screen for Task 4. So your prep window is used to convert observations into predictions, not to re-study the image.
Yes. The exact same image stays on screen. Task 3 asks you to describe what is happening now. Task 4 asks you to predict what will happen next or in the future. Smart candidates treat Tasks 3 and 4 as one connected exercise.
Future tenses. Mix ‘will’, ‘going to’, ‘might’, ‘could’, ‘is likely to’. Avoid using only ‘will’ for every prediction. Variety in speculative language directly raises your Vocabulary and Listenability scores.
Two or three solid predictions with reasons. A single prediction repeated three different ways is weaker than two clear, distinct predictions each backed by what is visible in the image.
No. There is no correct answer. Raters are listening for plausible reasoning expressed in clear future-tense English. A wild prediction with strong reasoning often scores higher than a safe prediction with no reasoning.
A brief 10-second mention is fine: ‘In this image, I can see a family packing boxes. They seem to be preparing to move.’ Then move straight into predictions. Do not re-describe the whole scene. That wastes your 60 seconds.
Avoid that. Use a range: ‘I think they will probably…’, ‘It’s likely that…’, ‘They might…’, ‘There is a good chance that…’. CELPIP raters specifically listen for future language variety in Task 4.
Tie every prediction to a visible clue. ‘Because they look stressed, I think they will…’, ‘Given that the sky is grey, it might…’, ‘Based on the boxes in the room, it seems they will…’. Visible evidence plus reasoning earns top-band Content scores.