CELPIP Speaking Task 3 (Describing a Scene) — a system that fills 60 seconds

Speaking Task 3 is where a lot of test-takers freeze. You get one picture of a busy everyday scene — a park, a market, an office, a train platform — and you have to describe it in detail to someone who cannot see it. The image is packed with people and small actions, so the problem is almost never 'what do I say' — it's 'how do I organise 60 seconds so I don't dry up at the 35-second mark and start repeating myself.'

This guide gives you a fixed system that works on any Task 3 image, the exact vocabulary that lifts your band, and a full sample response you can copy the rhythm of. Master it and you stop staring at the picture hoping words appear.

The Task 3 format at a glance

• You see one photo of a busy scene with several people doing different things.
• You get 30 seconds to prepare and 60 seconds to speak.
• Your job is to describe the scene to someone who cannot see it — be their eyes.
• You record once. There is no retry, so a plan matters more than perfect words.
• The same scene often reappears in Task 4 (Making Predictions), so read the image closely — it pays off twice.

What examiners score

Task 3 is rated on four criteria on the CELPIP 1–12 scale. Knowing what each one rewards changes how you speak.

• Content / Coherence — do you give enough relevant detail, in a logical order a listener can follow? A random list of objects scores low here even if every word is correct.
• Vocabulary — precise, varied word choice. 'A woman in a red coat is browsing a fruit stall' beats 'there is a woman.'
• Listenability — smooth, natural delivery a listener can follow easily: steady pace, few long pauses, clear linking.
• Task Fulfillment — did you actually describe the scene for someone who cannot see it, and use the full time well?

Notice that two of the four criteria reward organisation and listener-awareness, not vocabulary. That is why a system beats a bigger word list.

A system that never runs dry

Run the same five steps on every image, in this order. It guarantees you always know what sentence comes next, which kills the panic pause.

1. Overview — one sentence: what the scene is and where it is. 'This looks like a busy outdoor market on a sunny weekend morning.'
2. Foreground — the people closest to the camera and what each is doing. Give two or three of them a distinct action.
3. Background — the people and activity further back, behind or beyond the foreground group.
4. Details — weather, colours, and mood. These sentences are pure padding you can always fall back on when you blank.
5. Closing impression — one sentence on the overall feeling. 'Overall it feels lively and relaxed, like a normal Saturday.'

The genius of the order is that steps 4 and 5 are inexhaustible. If you reach 45 seconds and your actions are spent, you pivot to colours, weather, and mood and glide to the buzzer without a dead gap.

Describe by location, not randomly

The single biggest fix for a rambling answer is to move through the image on a fixed spatial path instead of jumping to whatever catches your eye. Pick a route and narrate it.

• Sweep in a consistent direction: foreground first, then background — or left to right. Never bounce around.
• Anchor every person to a place: 'on the left,' 'in the middle,' 'to the right,' 'in the background,' 'in the foreground.'
• Lean on prepositions of place: next to, behind, in front of, between, beside, near, at the edge of.
• Group people who are together: 'a family near the entrance' is faster and clearer than four separate sentences.
• Use the spatial links as your transitions — 'behind them,' 'further back,' 'off to one side' — so the answer flows.

Sample band-9 response

Here is a full 60-second answer for a busy park-market scene. Notice the spatial route, the present continuous, and how it leans on details near the end to fill the time.

This looks like a busy outdoor market in a city park, on a bright, sunny weekend morning. In the foreground, on the left, a woman in a red jacket is browsing a fruit stall and holding up an apple to check it. Right next to her, a young boy is tugging her sleeve, pointing at a stand selling ice cream. In front of them, a vendor in a green apron is arranging vegetables and smiling at a customer. Behind that group, in the middle of the scene, two friends are sitting on a wooden bench, sharing coffee and laughing. Further back, near the trees, a man is walking his dog on a lead, and a couple is taking a photo of a fountain. In the background, I can see more stalls with colourful striped umbrellas, and a few people are queuing for food. The sky is clear and blue, the trees are green and full, and everyone looks relaxed and cheerful. Overall, it feels like a lively, happy Saturday morning where people are out enjoying the sunshine.

That answer names roughly eight people or groups, gives each a distinct present-continuous action, moves cleanly from foreground to background, and closes with weather and mood. It comfortably fills 60 seconds without repeating a single action.

Vocabulary that lifts the band

You do not need rare words. You need three reliable families of everyday words, used precisely.

Present continuous verbs (the engine of the answer)

• is holding, is browsing, is pointing, is arranging, is waiting, is queuing
• is walking, is chatting, is laughing, is carrying, is looking at, is reaching for
• The whole task is happening now, so almost every action verb should be '-ing.'

Prepositions of place (the map)

• in the foreground, in the background, in the middle, on the left, on the right
• next to, behind, in front of, between, beside, near, at the edge of, across from

Adjectives (the colour)

• People: cheerful, relaxed, busy, focused, excited, elderly, young
• Scene: crowded, lively, bright, colourful, sunny, spacious, tidy
• Attach adjectives to the people you already named — 'a cheerful vendor,' 'an elderly man' — rather than adding new sentences.

What kills your score

• Listing objects at random — 'there's a tree, a bench, a dog, a car' — with no people and no order. This tanks Content/Coherence.
• Vague placeholders: 'there is a man,' 'there is a woman,' 'there are some people.' Add a colour, a place, and an action.
• Running out at 35 seconds and then repeating yourself or going silent — this is why you keep details in reserve.
• Describing what people are thinking or predicting the future — that is Task 4, not Task 3. Stay in the present, on what you can see.
• Speaking so fast to fill time that Listenability drops. Steady beats rushed.
• Forgetting your listener cannot see the image — always describe as if guiding someone with their eyes closed.