How Listen and Repeat is scored
Listen and Repeat is a Delivery-only task. Unlike the Interview or Independent Speaking tasks, it does not score you on Language Use (your grammar and word choice are fixed by the prompt) or Topic Development (there is no content to organise). Every point you earn or lose lives inside the Delivery dimension. That means raters are listening for five things on each item: lexical accuracy (did you say the actual words, in the actual order, with no drops or substitutions); prosodic match (did your stress and intonation contour track the model audio); pronunciation (are individual phonemes, especially consonant clusters and reduced vowels, intelligible to a non-strained listener); fluency (are you smooth and unhesitating across the 7-second window); and audibility (is your voice loud and clear enough for the microphone to capture cleanly). A response that hits all five at native or near-native level scores 5 on the task scale. Drop one criterion and you settle around 4. Drop two and you are at 3. The work of moving up a band, then, is rarely about a single dramatic improvement. It is about closing the smallest of the five gaps you currently have. If you want a fuller refresher on the task itself before reading on, see our Listen and Repeat strategy walkthrough.
The 5 source sentences
These are the target sentences. Each one is between 10 and 18 syllables, sounds plausibly academic or campus-flavoured, and contains at least one tricky cluster, reduced vowel, or unexpected stress placement. Read them aloud once before you scan the band attempts below.
- 1. The lecture starts at exactly nine o'clock on the third floor of Hayes Hall.
- 2. Researchers gathered statistical evidence that supports the original hypothesis.
- 3. If you submit your application by Friday, you will be considered for early admission.
- 4. Most species of migratory birds return to the same breeding grounds year after year.
- 5. The committee decided to postpone the meeting until further notice.
Band 3.0 attempt
A Band 3 candidate produces something that a sympathetic listener can mostly piece together, but each sentence carries at least one dropped or substituted word, one mispronounced cluster, and visible hesitation. Stress lands on grammatical particles instead of content words. The microphone often picks up trail-off at sentence ends.
"The lecture starts at... uh, exactly nine o'clock on the... the floor of Hayes Hall."
Rubric note. Drops "third"; pace is correct but stress lands on "starts" instead of "nine"; brief filler "uh" breaks the prosodic line and costs a fluency point.
"Researchers gather statistic evidence that support the original... hypotheses."
Rubric note. "Statistical" reduced to "statistic"; subject-verb agreement is altered ("support" for "supports"); final word substituted with the wrong number form. Three lexical errors in one sentence places this attempt firmly inside the 3 band.
"If you submit your application by Friday, you will... you be considered for early admission."
Rubric note. A self-correction mid-clause and the dropped "will" auxiliary in the second half flatten the conditional structure. Pronunciation of "considered" is intelligible but stressed on the wrong syllable (CONsidered instead of conSIDered).
"Most species of migrate birds return to the same... breeding ground year after year."
Rubric note. "Migratory" simplified to "migrate", because the candidate could not produce the four-syllable adjective under time pressure. Singular "ground" for plural "grounds" is a minor lexical drop, but layered with the earlier substitution it pulls the whole item below the 4 line.
"The committee decide to postpone the meeting until... further notice."
Rubric note. "Decided" is reduced to "decide" (past-tense "-ed" dropped). The pause before "further" is long enough that audibility briefly falls off; the final two words are clear but the rhythm of the sentence has already collapsed.
Band 5.0 attempt
A Band 5 candidate hits every word, in order, with intelligible phonemes throughout. What separates this tier from the ceiling is prosodic: stress sometimes lands on the wrong syllable of a multisyllabic word, and intonation contours are flatter than the model. Fluency is steady. Audibility is good.
"The lecture starts at exactly nine o'clock on the third floor of Hayes Hall."
Rubric note. All words present and intelligible. Stress on "EXactly" is slightly heavy where the model places it lighter; "third floor" is run together as one beat instead of two. Fully scoring on lexical accuracy but slightly under on prosodic match.
"Researchers gathered statistical evidence that supports the original hypothesis."
Rubric note. Words are correct. "Hypothesis" lands with stress on the third syllable (hy-po-THE-sis) instead of the second (hy-PO-the-sis), which is a clean tell of the 5 band. The sentence parses but the rhythm tells the rater you are reproducing words rather than melody.
"If you submit your application by Friday, you will be considered for early admission."
Rubric note. Lexically perfect. The conditional comma pause is held a beat too long, which slightly disrupts the fluency criterion. "Admission" is articulated cleanly. Stable 5.
"Most species of migratory birds return to the same breeding grounds year after year."
Rubric note. "Migratory" is articulated with all four syllables but stress lands on the second instead of the first (mi-GRA-to-ry instead of MI-gra-to-ry). The repetition "year after year" is rendered as three equal beats rather than the falling cadence of the model.
"The committee decided to postpone the meeting until further notice."
Rubric note. Every word in place. "Postpone" carries stress on the first syllable (POST-pone) where the model stresses the second (post-PONE). Audibility and fluency are clean. A small prosodic slip is the only thing keeping this from a 6.
Band 6.0 attempt
A Band 6 attempt is indistinguishable from a fluent speaker repeating the prompt. Every word is present, every stress lands where the model places it, intonation rises and falls with the source audio, and there is no audible processing time. You can hear the candidate listening to the melody, not the words.
"The lecture starts at exactly nine o'clock on the third floor of Hayes Hall."
Rubric note. Stress on NINE and THIRD, reduced "of" to a schwa, "Hayes Hall" delivered as a compound noun with primary stress on Hayes. Prosodic line matches the model end to end.
"Researchers gathered statistical evidence that supports the original hypothesis."
Rubric note. "Statistical" delivered with secondary stress on the first and primary on the third (sta-tis-TI-cal pattern). "Hypothesis" stressed correctly on the second syllable. Smooth across the longest source sentence on the page.
"If you submit your application by Friday, you will be considered for early admission."
Rubric note. Conditional rise on the comma break, falling resolution on "admission". "Will be" reduced naturally to a single weak beat rather than two clear words. This is the rhythmic detail that pushes a 5 to a 6.
"Most species of migratory birds return to the same breeding grounds year after year."
Rubric note. "MIgratory" stressed on the first syllable. Final phrase delivered with the natural English cadence of stressed-weak-stressed (YEAR after YEAR), not three equal beats.
"The committee decided to postpone the meeting until further notice."
Rubric note. "PostPONE" stressed correctly on the second syllable. "Until further notice" delivered as a single rhythmic unit with a falling tail, exactly mirroring the model. No filler, no hesitation, full audibility.
Side-by-side rubric breakdown
| Delivery sub-criterion | Band 3.0 | Band 5.0 | Band 6.0 |
|---|---|---|---|
| Lexical accuracy | Drops or substitutes 1 to 2 words per item; tense and number forms are unstable. | Every target word present and in correct order across all 5 items. | Every word present plus natural reductions on function words ("will be", "of the"). |
| Prosodic match | Stress lands on grammatical particles; intonation contour is flat or inverted. | Stress sometimes shifts on multisyllabic words but contour is roughly correct. | Stress and intonation track the model end to end, including question rises and resolution falls. |
| Pronunciation | Consonant clusters reduced or epenthesised; some phonemes unintelligible. | All phonemes intelligible to a non-strained listener; minor accent features preserved. | Native or near-native phoneme accuracy on every target. |
| Fluency | Visible hesitation, fillers ("uh"), at least one self-correction per item. | Steady pace with one or two long pauses across the 5 items. | No fillers, no audible processing time, smooth across all 5 items. |
| Audibility | Volume drops at sentence ends; final words sometimes inaudible. | Even volume across the response. | Even volume with confident projection on stressed beats. |
If you want to see this rubric mapped onto a different task type, the Band 3 Interview sample, the Band 5 Interview sample, and the Band 5 place-that-shaped-you sample show the same scoring logic applied to spontaneous speech.
What pushes you up the band ladder
- Shadowing. Pick any 60-second clip of a TED talk or an English news broadcast, and repeat each sentence within a 7-second window after the speaker. You are training the same verbal short-term memory the task taxes. The 2026 Listen and Repeat drills guide has a structured 14-day shadowing routine.
- Recording yourself daily. Pick 5 sentences each morning, record your attempt, then play it back against the source. Most learners can hear their own stress errors when listening at one remove, even if they could not feel them while speaking.
- Slow-then-fast drills. Read each target at half speed first, paying attention to the position of every stressed syllable. Then deliver it at full speed. The slow pass burns the prosodic shape into muscle memory.
- Attending to stress, not just sounds. Most candidates plateau at the 5 band because they are still listening for the words rather than the rhythm. Train yourself to hum a sentence on a single vowel before repeating it. If the hum matches, the words will land in the right place. The 2026 Speaking tips guide covers this drill in more depth.
What does NOT improve your band score
- Speaking faster. The 7-second window is generous for every sentence on this page. Rushing to fit more in produces dropped consonants and uneven stress, which costs you on three of the five sub-criteria. Pace your delivery to the model, not to the timer.
- Mimicking an American or British accent. Raters do not score for accent variant. They score for intelligibility, stress placement, and rhythm. A candidate with an unmistakable Indian or Brazilian accent who hits every stress can earn a 6. A candidate with a polished accent who fluffs the stress pattern earns a 5.
- Over-projecting volume. Audibility is a floor criterion, not a ceiling. Once your voice is loud and clear enough for the microphone, additional volume does nothing for your score and often degrades pronunciation by stiffening the articulators. Test your level with the practice mic, then deliver at conversational loudness.
Keep going
If you want more context on the 1 to 6 scoring system, the band-to-CEFR mapping, or the Speaking task structure, these companion pages will help.
- Listen and Repeat strategy walkthrough with three more worked examples.
- 2026 Listen and Repeat drills for a 14-day shadowing routine.
- 2026 Speaking Interview guide for the spontaneous speaking task.
- 2026 Speaking tips across all task types.
- How the 1 to 6 scoring system works.
- TOEFL band scores mapped to CEFR levels.
Drill these on a real test
Take a free TOEFL Speaking practice test with 7 Listen and Repeat items in the live-test format and 7-second timing, or sit a full mock exam for end-to-end timing pressure.
Try a Speaking practice test