PTE Repeat Sentence 2026: Strategy and Practice

🦈 Repeat Sentence is scored on fluency and pronunciation as well as words — and you cannot fix those by memorising sentence lists. SpeakShark gives you daily spoken-English conversation with phoneme-level feedback while you speak, plus an AI teacher that remembers the sounds you keep missing between sessions. Free tier, no card. Start a free speaking session →

Repeat Sentence is a working-memory task wearing a speaking task's costume, and it is the highest-leverage item in the PTE speaking section: Pearson's official Test Taker Score Guide lists 10 to 12 Repeat Sentence questions per test, more than any other spoken item type, and each one feeds four separate score lines — Listening, Speaking, Pronunciation and Oral Fluency. You cannot brute-force it by memorising sentence banks, because the sentences are new. What you can train is the mechanism: hearing English in meaningful chunks instead of individual words, and starting to speak instantly even when you only caught seventy percent. This guide uses Pearson's published scoring criteria, explains what the memory science says about why nine-second sentences feel impossible, and gives you 20 original practice sentences graded by length.

In this guide: at a glance · what it really tests · how it is scored · the chunking strategy · what to do when you miss words · 20 original practice sentences · a 14-day plan · where the sources disagree · when this is the wrong prep · how we researched this · FAQ

Repeat Sentence at a glance

Two numbers in that table decide how you should budget your prep time. 10-12 items means Repeat Sentence outnumbers Read Aloud (6-7), Answer Short Question (5-6), Describe Image (5-6) and Retell Lecture (2-3). And Listening and Speaking means every one of those items is counted twice — once toward each communicative skill. If your Listening score is dragging your overall down, Repeat Sentence is probably where it is leaking, not the Part 3 listening section.

For the wider picture of how the speaking tasks fit together, start with our PTE Academic speaking overview, then work through the task guides: Read Aloud, Describe Image, Retell Lecture and Answer Short Question.

What Repeat Sentence really tests: working memory, not vocabulary

Here is why the nine-second sentences feel physically impossible even when you understood every word.

Verbal short-term memory is not a tape recorder. In the standard model, incoming speech lands in a phonological store that holds it briefly and then loses it. A 2024 review in the Quarterly Journal of Experimental Psychology describes the classic version of that store as holding "representations of verbal items in phonological form for around 2 s before they are lost to decay" (Hughes, 2024). Roughly two seconds. A Repeat Sentence prompt runs up to nine.

The gap is closed by two things, and only two:

• Rehearsal — the inner voice that recycles the sound back into the store. This is why the store's practical capacity tracks how fast you can articulate, not how many words you heard. It is also why rehearsal collapses when you are simultaneously translating.
• Chunking — recoding many small items into fewer meaningful ones. Capacity research converged on a limit of about four chunks in adults rather than Miller's classic seven (Morra, Patella and Muscella, 2024, Journal of Cognition, summarising Cowan's estimate that "the limited focus-of-attention capacity averages about four chunks in normal adults").

Four chunks. A 14-word sentence is 14 items if you hear it as words — three or four items if you hear it as phrases. That single recoding step is the whole difference between candidates who reproduce long sentences and candidates who reliably lose the back half.

Pearson's own guidance points at exactly this. In its speaking self-improvement article, Pearson tells test takers to comprehend the meaning to aid retention, to use "English rhythm with chunking and sentence stress," and — notably — to keep the sentence in your head rather than reaching for the erasable noteboard. Scribbling costs you the rehearsal loop you need.

The practical consequence: if you are translating the sentence into your first language while it plays, you are using the same articulatory machinery that should be rehearsing it, and the sentence will be gone before the microphone opens. Repeat Sentence rewards people who process English directly. That capacity is built by hours of listening to and producing natural-speed English — not by sentence lists.

How Repeat Sentence is scored: three traits, four score lines

Pearson publishes the exact band descriptors. Here they are, condensed from the Test Taker Score Guide.

Content (0-3). Pearson defines errors narrowly: "Errors = replacements, omissions and insertions only," and adds that "hesitations, filled or unfilled pauses, leading or trailing material are ignored in the scoring of content."

Pronunciation (0-5) and Oral Fluency (0-5) are then applied, but only if Content is above zero. The score guide's worked example spells out the gate: if the Content score is greater than 0, "the response will be rated on the two remaining traits: Pronunciation and Oral Fluency."

The Oral Fluency descriptors are where most candidates quietly lose points, so read these closely:

• 5 Highly proficient — "Speech shows smooth rhythm and phrasing. There are no hesitations, repetitions, false starts or phonological simplifications."
• 4 Advanced — "no more than one hesitation, one repetition or a false start."
• 3 Good — "Speech is at an acceptable speed but may be uneven... There are no long pauses and speech does not sound staccato."
• 2 Intermediate — "Speech may be uneven or staccato. Speech (if >= 6 words) has at least one smooth three-word run, and no more than two or three hesitations, repetitions or false starts."
• 1 Limited — "irregular phrasing or sentence rhythm... multiple hesitations, repetitions, and/or false starts."
• 0 Disfluent — "Most words are isolated, and there may be more than one long pause."

Notice what band 2 measures: whether you can produce one smooth three-word run. Not whether you got the words right. Fluency here is literally scored on continuous phrase runs — which is the reason chunking is not a memory trick but a scoring strategy.

Pronunciation is scored on intelligibility, not accent. Band 5 requires that "all vowels and consonants are produced in a manner that is easily understood by regular speakers of the language," with appropriate assimilation and correct word and sentence stress. Band 3 — "Good" — still allows "some consistent errors" that "might make a few words unclear."

That is a trainable target, and it is exactly what conversational practice with real-time phoneme feedback moves: not your accent, but the handful of consonants and stress patterns that make a listener stumble. Try a free session and hear which sounds get flagged →

The chunking strategy that actually moves the score

Chunking is the one technique that raises Content and Oral Fluency at the same time. Here is the method, in the order you should run it.

1. Listen for meaning, not words. From the first syllable, ask what the sentence is about. Meaning is the glue that holds a 14-word string together; a list of 14 unrelated sounds has no glue at all.

2. Break the sentence at its natural phrase boundaries. English speech already comes pre-grouped — noun phrases, verb phrases, prepositional phrases — and native speakers mark those groups with tiny pitch resets and micro-pauses. Take this original example:

Although the sample size was small, / the findings were consistent / across all four groups.

Fourteen words, three chunks. Three is inside your four-chunk budget. Fourteen is not.

3. Rehearse silently in the gap. Between the end of the audio and the microphone opening there is a short beat. Use it to run the chunks once in your inner voice, in order. Do not use it to worry.

4. Speak the chunks as chunks. Deliver each group as one continuous run with a single stressed word, then a clean boundary. This is what earns the "smooth three-word run" the Oral Fluency descriptors are looking for, and it is why word-by-word delivery scores badly even when every word is correct.

5. Start immediately. Pearson's format page is explicit that "once the microphone shows as recording, you will need to speak straight away." A long silence at the front is both dead air and a hesitation.

6. Copy the intonation, keep your accent. Pearson advises mirroring the speaker's intonation while preserving your own accent. Sentence stress carries meaning in English; flattening it costs pronunciation points even when every phoneme is right.

The counterintuitive part: do not slow down to be careful. Slow, deliberate, word-by-word repetition is exactly the profile the Oral Fluency band descriptors call "staccato" and "isolated." A natural conversational rate with clean phrase groups outscores a careful crawl, every time.

What to do when you miss words

This is the moment that decides most Repeat Sentence scores, and most candidates handle it wrong.

Never stay silent. Content 0 means "almost nothing from the prompt in the response" — and a Content of 0 also kills Pronunciation and Oral Fluency for that item, because those traits are only rated when Content is above zero. One panicked silence costs you three score lines on an item that appears 10 to 12 times.

Say half and get 2 out of 3. The band descriptor is unambiguous: 2 points for "at least 50% of words in the response from the prompt in the correct sequence." Half a sentence, delivered fluently, is two-thirds of the available Content score plus full eligibility for Pronunciation and Oral Fluency. That is a strong item.

Guess out loud rather than stopping. Pearson's own advice on missed words: if you are unsure of a word the speaker used, say what you think you heard. A wrong guess is one replacement error. A pause while you hunt for the right word is a fluency penalty and usually loses the rest of the sentence too.

Keep the original order. Sequence is part of the Content definition at every band. Words delivered out of order do not count the same as words delivered in order.

Do not repair. If you realise mid-sentence that you said "were" instead of "was," keep going. Going back is a false start and a repetition — two of the three things the Oral Fluency descriptors count against you.

Do not pad. Leading and trailing material is ignored for Content, so filler adds nothing there, and it adds hesitation risk to your fluency score. Say the sentence, stop, wait.

A useful way to hold all six rules at once: your job is not to be accurate, it is to be fluent about whatever you managed to catch. If Repeat Sentence is one of several tasks dragging your result down, our guide to why your PTE speaking score is low walks through the other common leaks.

20 original practice sentences, graded by length

These are original sentences written for this guide — they are not real PTE items and are not reproduced from any exam, item bank or other site. They are modelled on the academic register and the 3-to-9-second prompt length Pearson publishes.

Use them like this: have someone read one aloud once at normal speed, or record yourself reading them and play them back cold days later. Never read them while repeating.

Short (about 3-4 seconds) — build the reflex of starting instantly:

Medium (about 5-6 seconds) — practise breaking into two chunks:

Long (about 7-9 seconds) — three chunks, and the real test:

How to mark yourself. Write down what you said, compare to the original, and count only replacements, omissions and insertions — the same three error types Pearson counts. Then listen back to the recording once more, ignoring the words entirely, and ask a single question: did that sound like continuous phrases, or like a list?

A 14-day Repeat Sentence practice plan

Fifteen to twenty minutes a day. The plan is built around the two things the scoring actually measures — chunk-sized listening and continuous delivery.

Days 1-3: rebuild the listening channel. Listen to unscripted English at natural speed for 10 minutes daily — podcasts, lectures, interviews, anything unscripted. Then repeat five sentences from the short table above. The goal this week is not accuracy; it is not translating.

Days 4-7: chunking drills. Take the medium sentences. Before repeating, say the chunk boundaries out loud ("the deadline for coursework submission / has been moved to Friday"), then repeat the whole sentence as two runs. Add one long sentence per day at the end.

Days 8-11: pressure and recovery. Work only from the long table, and deliberately practise the failure case: have your partner mumble one word, or play the audio at slightly reduced volume, and force yourself to answer anyway inside 15 seconds. This is the single most valuable rep in the plan, because item 7 on test day will be one you only half-caught.

Days 12-14: fluency and sound. Record every rep. Listen back for hesitations, repetitions and false starts — the exact three items in Pearson's Oral Fluency descriptors — and for the words that come out mushy. Drill those sounds inside full sentences, never in isolation, because pronunciation is scored in connected speech.

Underneath all fourteen days there has to be a base layer of actual conversation, because that is what builds the automatic English processing Repeat Sentence measures. This is where a daily AI speaking session earns its place: SpeakShark's free tier gives you three 5-minute sessions a day with teacher Sarah, 64 topics, no card — and phoneme-level pronunciation feedback delivered while you are talking, not in a report afterwards.

The differentiator that matters for a 14-day plan is memory. SpeakShark's AI teacher carries your recurring mistakes, your level and your topics from one session to the next, so day 12 already knows what day 4 exposed. Every other AI practice app meets you as a stranger each morning, which means you spend the first two minutes re-establishing what you are working on. Premium — $10/month or $69/year, about $5.75/month, on the pricing page — adds unlimited sessions, 10-minute conversations with unlimited turns, 320 topics and all four teachers (American, British, Australian, Canadian), which matters if you want to practise against an accent other than the one you find easiest.

Where the sources disagree on timing

Being able to see a contradiction is more useful than a false consensus, so here it is.

Two of Pearson's own pages give different answer windows for the same task type in two different products. We are not going to pretend one of them is wrong. What we will say is that it changes nothing about your strategy: whether the timer says 15 or 40, you should begin speaking immediately and finish in one continuous pass, because dead air is scored against you either way. A longer window is not an invitation to think.

The score guide also carries a caveat worth remembering: Pearson notes that the minimum and maximum timings for each section "do not add up to the total timings stated," because "different versions of the test are balanced for total length," and that "no test taker will get the maximum or minimum times indicated." In plain terms — your test will differ from the published table. Trust the screen.

For context on the wider test, the Test Taker Score Guide (checked July 2026) states you will complete between 65 and 75 questions across 22 question types, that the overall score runs 10-90, and that the whole assessment is a maximum of 2 hours 15 minutes.

When Repeat Sentence practice is the wrong thing to do

An honest guide has to say where its own advice stops applying.

If you cannot follow natural-speed English at all, sentence drills are premature. Repeat Sentence is downstream of listening comprehension. Grinding sentence lists when you cannot yet parse conversational English trains frustration, not skill. Build the listening base first with unscripted audio and daily conversation, and get an honest baseline before you book a test date — our free English speaking level test is a starting point.

If your problem is one or two stubborn consonants, use a drill app. Tools built for isolated phoneme repetition with mouth-position graphics do a job conversational practice does not. For a narrow, specific sound problem, that is sharper than any conversation tool including ours.

If memorised sentence banks are your plan, stop. The sentences are new each test. Worse, Pearson explicitly classifies responses containing significant amounts of pre-prepared or memorised material as irrelevant, scoring Content as 0 on the task types where that applies. Memorisation is not just ineffective here — it is a documented risk.

If nerves are the bottleneck, fix the nerves. A candidate who freezes for four seconds has lost the item before language is even involved. Our guide to overcoming the fear of speaking English covers the graded-exposure approach that actually shifts this.

If your goal is a job rather than a visa, you may be preparing for the wrong test. The same Versant speech-processing engine that scores PTE speaking also powers Pearson's employer-facing Versant tests — see our guides to the Versant test at Accenture and how to prepare for the Versant test. And if the real target is a conversation with a human, structured English role-play scenarios will serve you better than any repetition drill.

How we researched this guide

Every factual claim above comes from a page or document we opened in July 2026, linked so you can check it.

Primary sources (Pearson, the test maker): the PTE Academic Test Taker Score Guide (Copyright Pearson Education Ltd 2025) for item counts, the Content 0-3 band descriptors, the full Pronunciation and Oral Fluency criteria, the Content-greater-than-zero gate, the list of human-reviewed question types, the 65-75 question total, the 10-90 scale and the statement that spoken responses are scored by Pearson's Versant technology; the PTE Academic Speaking and Writing format page and the PTE Core equivalent for prompt length and answer windows; the Part 1 introduction article for the maximum trait scores and the instruction to speak as soon as the microphone opens; and Pearson's speaking self-improvement article for its advice on chunking, keeping the sentence in your head, and saying what you think you heard.

Cognitive science sources, used for the working-memory explanation only: Hughes (2024) in the Quarterly Journal of Experimental Psychology for the roughly two-second phonological store, and Morra, Patella and Muscella (2024) in the Journal of Cognition for the four-chunk capacity estimate. These describe general verbal memory, not PTE performance specifically; we use them to explain why chunking helps, not to predict anyone's score.

What we deliberately did not do: reproduce real PTE items. Every one of the 20 practice sentences on this page was written from scratch for this guide. We also did not publish a "target score" for Repeat Sentence, because Pearson reports skill scores and an overall score, not per-task-type scores — any per-item target you see quoted elsewhere is an inference, not a published figure.

SpeakShark is a speaking-improvement tool, not an exam-prep provider, and is not affiliated with or endorsed by Pearson, PTE Academic, IELTS, or any exam board.

FAQ

Ten to twelve times on test day, a voice will say something once and then stop. The candidates who score well are not the ones with better ears — they are the ones who hear English in phrases instead of words, and who open their mouths immediately with whatever they caught. Both of those are reflexes, and reflexes are built by speaking English every day between now and your test date.

Start your free daily speaking session on SpeakShark → — three sessions a day, no card, with pronunciation feedback while you speak and a teacher that remembers what you got wrong yesterday.

All Pearson item counts, band descriptors, timings and policies verified July 2026 from the official documents linked above; Pearson updates these guides periodically, so confirm the current version before your test. The 20 practice sentences are original to this guide and are not PTE material.