A new speaking topic every day feels productive. It also forces you to invent content, recall vocabulary and organise sentences all at once. You may finish ten topics without ever hearing the same message become easier to deliver.
The 4/3/2 speaking drill deliberately removes the new topic. Give the same talk in four minutes, then three, then two. The shrinking time pushes you to cut repetition and long pauses while keeping the essential message.
What the original activity involved
Paul Nation described the technique in 1989. A learner prepared a talk and delivered it to three different listeners, for four, three and two minutes. The listeners did not interrupt or ask questions. Nation reported improvements in speaking rate, hesitation and some grammatical measures during the activity, but the study was small. Treat it as a promising task design, not a guaranteed numerical outcome.
Three features do the useful work:
- Repetition reduces the effort of deciding what to say.
- Less time creates gentle pressure to become more efficient.
- A new listener makes the repeated talk an act of communication rather than recitation.
When practising alone, change the imagined listener or purpose on each round to preserve the third feature.
A solo version that needs only four prompt words
Choose something you already know well: a problem you solved, your daily workflow, or the plot of a short video. Write four content prompts, not a full English script.
1. Four minutes: the first-time listener. Explain the background, sequence, result and your opinion. Record yourself. If a word is missing, describe around it instead of stopping. 2. Three minutes: the busy colleague. Keep the main message but combine repeated details. Make the order easier to follow. 3. Two minutes: the friend leaving in a lift. Start with the result, then give only the context needed to understand it.
Do not rewrite the talk or look up new vocabulary between rounds. You want to see what changes when the same resources become more accessible, not hide the first attempt behind a polished script.
Scale the clock to your current level
Four, three and two minutes belong to the original classroom setup; they are not a proficiency test. If you can currently speak for one minute, use 90/60/45 seconds. For a self-introduction, try 120/90/60 seconds.
Keep the three principles: familiar content, decreasing time and meaningful delivery. If the first round cannot fill the time, narrow the topic. "Why I chose this restaurant" is easier to sustain than "food culture in my country" without preparation.
Review three things, not every mistake
- Pauses: Are they between ideas, or after every second word while you search?
- Information: Did the last round remove side details, or did it lose the cause and result too?
- Repetition: Do *and*, *very* and *I think* connect every sentence? Choose one repeated habit to adjust.
Note the actual duration and one improvement for each round. Word count alone is a poor score: faster speech that becomes harder to understand is not the goal.
Four ways to turn it into empty repetition
Memorising a script first. Fluency practice should use familiar language and content. A memorised script changes the task into recall or reading aloud.
Adding harder material in the final round. The two-minute version should be tighter, not packed with new vocabulary.
Judging only the first attempt. The first recording is a baseline. A clearer final round with fewer long searches is useful evidence of change within the task.
Keeping the same talk forever. Finish the set and leave it. Retell it on another day to a different imagined listener if you want to test whether the improvement survives.
Where MirrorSay fits
After the three rounds, take the two sentences where you most often stalled. Say your intended meaning to MirrorSay, compare a natural formulation and listen to the model before repeating it aloud. Timed retelling works on the whole message; MirrorSay helps you repair the individual sentence.
For sentence-level rhythm, read English Thought Groups. To bring successful lines back on later days, use Spaced Repetition for Speaking.
If today’s version is fluent, use spaced task repetition to check whether you can reorganise it tomorrow without shrinking the time again.
References
- Nation, P. (1989). Improving speaking fluency. *System*, 17(3), 377–384. DOI · Author-hosted PDF
- Nation, P. (2007). The Four Strands. *Innovation in Language Learning and Teaching*, 1(1), 2–13. DOI · Author-hosted PDF
*MirrorSay is a free AI speaking-practice app: say a sentence, see a natural way to express it, then shadow the model. No sign-up, and speech is processed on-device whenever possible.*