PTE Core Speaking: lesson 4 of 7

Lesson 4 of 714 min

Repeat Sentence: Scoring and Memory

How Repeat Sentence content is scored word by word, when chunking a sentence actually helps, and how to tell a listening problem from a memory problem.

Repeat Sentence has one of the highest item counts in PTE Core, and it feeds both Speaking and Listening. This lesson explains exactly how content is scored, when splitting a sentence into chunks helps, and a simple test for whether your mistakes come from hearing or from remembering.

The task at a glance

Per Pearson's PTE Core Score Guide, there are usually 10–12 Repeat Sentence items. After a short pause, a sentence plays once; recording starts right after the audio ends, with no beep, so watch for the status to change and then speak. Check the latest official guide before your test.

  • Sentence length: typically about 9–14 words (a practice tip, not an official figure). Repeating one comfortably takes well under 15 seconds.
  • The 3-second rule: after about 3 seconds of silence the microphone closes (a widely reported test behaviour; Pearson's Score Guide does not state the figure). If you can't remember the whole sentence, say what you have. Don't stop to recall.
  • Two scores: Repeat Sentence counts toward Speaking and Listening, so weak listening affects it twice. That makes it worth prioritising.

The hard part isn't pronunciation, as in Read Aloud. It's short-term memory: you hear the sentence once, with no text to look at.

How content is scored

Per the Score Guide, Repeat Sentence Content has four bands:

ScoreDescriptor
3All words in the response from the prompt, in the correct sequence
2At least 50% of words from the prompt, in the correct sequence
1Less than 50% of words from the prompt, in the correct sequence
0Almost nothing from the prompt

The Score Guide defines errors as replacements, omissions and insertions only, and treats them equally. Saying "attracting" instead of "recruiting" costs the same as leaving the word out. It also states that hesitations, filled or unfilled pauses, and leading or trailing material are ignored for Content, though they still affect Fluency.

Three things follow:

  1. Word order matters. Every band says "in the correct sequence". Remember the sentence by its structure, not as a bag of words.
  2. A synonym is not a safe substitute. It's a replacement, which is an error. Say what you heard, not what you think it meant.
  3. Pronunciation and Fluency are scored separately. Use your own natural intonation. Copying the flat, mechanical delivery of the recording can pull those traits down.

Choosing a default approach (a practice tip). Decide before test day what you'll do when a sentence is too hard, so you're not deciding under pressure:

  • Most of it understood: skip the odd word you missed and deliver 80% or more, smoothly.
  • About half understood: give the half you're sure of (first half, second half, or beginning and end) and don't force the rest.
  • Very little understood: say the two to four content words you caught, fluently, rather than freezing.

Practice sentences often come in near-identical versions, for example "We are committed to recruiting the best students regardless of their financial situations" and "We are committed to attracting the very best students regardless of their financial circumstances." The scoring follows the words you actually heard, not the version you expected.

Chunking: when it helps and when it doesn't

Short-term memory holds only a few items at once. Fourteen separate words overload it; three or four meaningful chunks fit. That's the case for chunking. But it has limits (a practice tip based on teaching experience):

  • Up to about 12 words: don't bother chunking. A short sentence already fits in memory, and dividing it just adds work while you're trying to listen. Many Repeat Sentence items fall in this range.
  • Longer than about 12 words: chunking starts to pay off.
  • At any length, if you didn't hear the words clearly, chunking can't help. Chunking solves memory problems, not listening problems. Run the diagnosis in the next section before you invest time here.

When a sentence is long enough, group it into phrases: a verb phrase, a noun phrase, a prepositional phrase.

You need to prepare / your boarding pass and passport / in front of the gate.

Fourteen single words become three chunks, and your knowledge of grammar fills in the inside of each one.

Restricted scholarships / are targeted principally at / students with specific goals.

Two related habits help: split by meaning ("The government / has implemented / a new policy") and treat fixed phrases as one unit ("make a decision", not make + a + decision). In practice, divide the sentence silently as you listen; there's no time to write.

When you miss part of it:

  1. Let grammar fill small gaps. If you caught "Students ... available in the department", a natural linking word ("are") gets you further than silence. Keeping the beginning and end you did hear scores better than stopping.
  2. Start speaking. Waiting to be sure costs time, and after 3 seconds the recording closes.
  3. Use your own voice. Don't imitate the recording's accent or flat delivery.

Variations on one pattern. Practice sets often contain several sentences built on the same frame:

The bus for London will leave ten minutes later than planned. The bus for London will leave ten minutes earlier than planned. The bus for the airport will arrive soon. The bus by the building goes directly to the central bus station.

The frame is [The bus + description] / [will + verb phrase] / [time or place]. Once that frame is automatic, you only need to catch what fills each slot. But "later" and "earlier" sit in the same slot and mean opposite things. If you say the version you remember instead of the one you heard, it's a replacement.

Is it listening or memory?

This is the most important step in the whole Repeat Sentence section. The same result, "I didn't repeat it correctly", can come from two different problems with opposite fixes (a practice tip; the procedure is ours, not Pearson's):

  • Memory: you heard it but lost it halfway through. Practise chunking and separating listening from speaking (next lesson).
  • Listening: some words were a blur the moment they arrived. Chunking won't help at all. Train your ear instead.

How to tell them apart (on your own):

  1. Play a sentence once, then write what you heard. Take as long as you need. Spell unclear bits as they sounded.
  2. Only then compare with the original.
  3. You can write it correctly but couldn't say it all: memory problem. You can't write it, or you wrote wrong or garbled words: listening problem. Right words, wrong order: a memory problem with order.

Writing takes away the pressure of speaking, so any remaining gap must come from listening.

Do ten sentences and count each type. Whichever is more than half is what you practise for the next two weeks. Then test again.

Training your ear. Replaying audio you already can't decode doesn't fix it. Most listening misses come from three features of connected speech:

  • Linking: "pick it up" sounds like one word; "an hour" sounds like "a nower".
  • Reduction: function words shrink to /ə/. "I can go" has a tiny /kən/, while "I can't go" is stressed, which is why people hear positive sentences as negative.
  • Assimilation and elision: "would you" sounds like "wudju"; the t nearly disappears in "next day".

Four steps, 10–15 minutes a day, using sentences you've already practised:

  1. Dictation: replay one sentence and write it word by word until you can't add anything (usually five to eight plays). No peeking.
  2. Mark only the misses. This is your personal list of words you tend to mishear.
  3. Label each miss: linking, reduction, or assimilation and elision. The category matters more than the individual word, because it recurs across dozens of sentences.
  4. Shadow the sentence half a second behind the audio, linking and reductions included, five times. Linking you can say is linking you'll recognise.

After two weeks, run the diagnosis again. If the listening misses haven't dropped, check your labels. Often an "unheard" word was actually an unknown word, and that's a vocabulary job.

Rating your own Content score

Repeat Sentence is the only speaking task where you can work out your Content band exactly. For each sentence:

  1. Count the words in the original: N.
  2. Count replacements, omissions and insertions in your recording: E.
  3. Check the order of the words you did repeat.
  4. Leave hesitations and extra filler out of E, but note them for Fluency.

Then: E = 0 and correct order → 3. Words from the prompt in correct order at least N/2 → 2. Fewer than N/2 → 1. Almost nothing → 0.

Notice the jump. Going from zero errors to one drops you a full band; going from one to four often changes nothing, as long as you stay above half. The first error is the expensive one. That's why the practice check below counts error-free sentences, not average errors.

Worked examples, chunked:

  1. Restricted scholarships / are targeted principally at / students with specific goals.
  2. You must wear a hard hat / when you go to the construction site.
  3. Parking permits can be collected / through the student service office. (Don't turn "can be collected" into an active verb; that changes several words.)

For seven more sentences with chunks and typical errors, see the practice sentence set, then try them in Repeat Sentence practice.

Practice checks: first, run the listening-or-memory diagnosis. Then take ten practice sentences and rate each one: aim for at least eight at band 3 (zero errors, correct order) and the other two at band 2, with no silence longer than a second. Keep the one-page Repeat Sentence Memory Chunks open while you practise.

Sources

Reviewed by PrepAmigo editorial team

Put it into practice

Practice Repeat Sentence

Start practice