PTE Academic · 4 min read

Repeat Sentence: Hold the Whole Utterance, Then Say It Back

How Repeat Sentence is scored on content, oral fluency and pronunciation, with a chunking method, a worked academic example, and the hesitation trap that costs more marks than a missing word.

Repeat Sentence checks whether you can hold a short spoken utterance in working memory and reproduce it as speech, not whether you understood its meaning in depth. The automated scorer listens for three things at once: how much of the original wording you reproduce, whether your delivery runs at a smooth, even pace, and whether your sounds and stress are recognisable to an English listener. Because all three contribute, a half-remembered sentence delivered confidently usually outperforms a nearly complete one delivered in stop-start fragments.

A method you can repeat under pressure

Before the audio starts, stop rehearsing the previous item and look away from the screen; divided attention is the main cause of lost words.
Listen to the whole sentence without trying to transcribe it mentally — attempt to hold it as two or three meaning chunks, such as subject group, verb group and final detail.
Echo the rhythm internally: note where the speaker's voice rose or fell and which words were stressed, because that contour helps you recover wording.
Begin speaking promptly once the recording ends and keep one continuous pace; do not restart a phrase you have already produced.
If a middle word is gone, replace it with a plausible word of the same shape and keep the sentence moving rather than stopping to search.
Finish the final chunk with falling intonation so the response sounds like a completed sentence, then stop talking.

Worked example

These examples explain the difference. Use your own words for each prompt.

Do this: Prompt heard: "The lecture on coastal erosion has been rescheduled to Thursday afternoon." Response: "The lecture on coastal erosion has been rescheduled to Thursday afternoon." — delivered in one pass at an even pace, with stress on "coastal erosion" and "Thursday", and a clear falling tone at the end.

Not this: Response: "The lecture on... on coastal, um, coastal erosion has been... re-, rescheduled... to Thursday. Thursday afternoon." — three restarts, two filled pauses, and the final phrase tacked on after a long gap.

Both responses contain almost the same words, yet they score very differently. The first keeps content intact and delivers it as a single rhythmic unit, so oral fluency and pronunciation ratings stay high; stressed words land where a native speaker would place them, which also makes individual sounds easier to recognise. The second shows the same recall, but every repetition, filled pause and restart is counted as disfluency, and the broken rhythm flattens the stress pattern, making words such as "rescheduled" harder for the scorer to identify. The lesson is that recall and delivery are scored together — protecting delivery protects the other two dimensions.

The common mistake

The most damaging habit is treating a forgotten word as a reason to stop. Test takers freeze mid-sentence, say "sorry", or begin again from the start, and lose far more than the single word that was missing. A partial sentence spoken fluently from beginning to end still earns content credit for everything you produce; an abandoned attempt earns very little and leaves silence where speech should be. A second, quieter mistake is starting to speak before the audio has fully finished, which clips the ending you most need to hear.

Build the habit of deliberately completing damaged sentences: take a prompt, intentionally drop one word, and say the rest smoothly anyway. Training yourself to continue through a gap turns a memory lapse into a minor content loss rather than a collapsed response.