Making Predictions

This task does not ask you to describe the scene. It asks you to reason forward. Notice what is already happening in the scene, then say what will probably follow and why. A strong answer links each prediction to something visible in the scene, so a listener who cannot see it can still follow your thinking. Description is useful only as a short setup for your predictions.

Step by step

1. 1

   Scan for unfinished actions

   When preparation starts, look for things that are still in progress: someone reaching for something, an object close to falling, a crowd gathering, a sky turning dark, a face showing worry or excitement. Unfinished actions point straight to what comes next, so they give you ideas faster than listing everything you see.

2. 2

   Pair each prediction with its clue

   Choose the developments you feel surest about, and for each, pick the visible detail that supports it. Making these pairs now means every claim you make later has a reason behind it, and you never have to invent a reason while recording.

3. 3

   Set a route through your ideas

   Decide on a simple order and keep it in your head as a short chain of key words. You could move from what happens soon to what happens later, from the main people to the people in the background, or from a cause to its effects. A route stops you from jumping around. If the test allows notes, write key words only, never full sentences.

4. 4

   Frame the scene briefly, then move to the future

   Begin with a short orientation so the listener knows where they are and who is involved. Then turn to the future quickly. The only job of the description is to set up your first prediction.

5. 5

   Develop each prediction with a reason and a detail

   For each idea, say what will happen, why you expect it, and a concrete detail about how it plays out, such as how a person reacts or what changes in the setting. Show how sure you are: some outcomes are almost certain and others are only possible, and marking the difference makes your reasoning sound genuine.

6. 6

   Land on an outcome

   Near the end, bring the scene to a resolution: how things turn out for the people involved. If you sense time running short, skip ahead to this ending so your answer finishes on a complete thought instead of stopping mid-idea.

Using the time

At the start, use the preparation time to find unfinished actions and choose your clue and prediction pairs. Do not spend it polishing sentences. Once recording begins, keep the framing short and give most of the time to developing predictions with reasons and details. Save the end for your outcome. Keep a rough sense of how much time has passed. If you are still on your opening idea when you feel the time running out, cut it short and move to the ending.

What usually goes wrong

• Describing the present instead of predicting

  It is easy to keep talking about what people are doing now because it is right in front of you. Listen for yourself: if you have said several sentences about the present in a row, switch to what happens next at once.

  In your feedback: Off prompt

• Predictions with nothing behind them

  A dramatic outcome with no link to the scene sounds like a random guess. After each prediction, silently ask yourself why. If you cannot point to a visible detail, add one or replace the prediction with one you can support.

  In your feedback: No supporting example

• Running dry and going silent

  If you use up your planned ideas early, you may freeze. Keep a spare idea ready, such as a background person or a later consequence of your main prediction. Move to it as soon as you feel a gap opening instead of waiting in silence.

  In your feedback: Long pauses

• The same likelihood word again and again

  Predictions invite one future word used over and over. When you hear yourself repeat it, change how you mark likelihood or certainty. Varying these words also shows more clearly which outcomes you are sure of.

  In your feedback: Repetition

Checking your own work against what is judged

Content/Coherence

Content/Coherence: Check that each prediction came with a clue from the scene and a concrete detail, and that your ideas followed the route you chose, from a brief setup to an ending.

Vocabulary

Vocabulary: Check whether you named objects, actions and feelings precisely instead of using general words, and whether you varied the words you used to show how likely each outcome is.

Listenability

Listenability: Ask whether someone who never saw the scene could follow you. Also check for fillers or silences between ideas, especially at the moments when you moved from one prediction to the next.

Task Fulfillment

Task Fulfillment: Check that most of what you said was about the future, not the present, that you dealt with the main people or action in the scene, and that you reached an outcome before recording stopped.

Drafted by an AI writer and passed by a separate AI editor after one rewrite, checked against our content rules: structure and approach only, no wording to reuse. How our content is made