Listen and Repeat, explained
You hear a sentence once, you never see it written, and you say it back. There is no question to answer and no opinion to give. This page explains what the scorer is comparing, which words go missing first, and prints six sentences you can rehearse from the page.
What the Listen and Repeat task is
Listen and Repeat is one of the two Speaking task types on the test that took effect on January 21, 2026. It measures how much of a spoken sentence you can hold and reproduce — a narrower thing than "speaking ability", and a much more trainable one.
What ETS states
The 2026 specification lists the Speaking section at 11 items and 55 raw points, with task types Listen and Repeat (7 items) and Take an Interview (4 items).
The same specification table gives the maximum points for each Listen and Repeat and Take an Interview item as 5, which is how 11 items reconcile to 55 raw points. The specification states that its published blueprint is subject to minor revisions until the official launch.
All Speaking tasks are AI-scored constructed responses, per the specification’s item-type notes.
This describes how ETS states the test is scored, and says nothing about how Reach120 scores your practice.
Sourced from the ETS 2026 test specifications, read (link confirmed ). Everything outside this box is Reach120’s own reading of it.
What the specification does not publish is a per-item time limit, a required sentence length, or the sentences themselves. Anywhere you see those stated as rules, you are reading a guess. The tiers and sentence sets on this page are Reach120 practice settings chosen to build the skill; they are not ETS rules.
What the scorer is actually comparing
This is the one speaking task where the correct answer is known in advance. Your recording is transcribed and the transcript is compared with the sentence that was played — no model is deciding what you should have said. Two things are scored:
| What is scored | What it measures | What a strong response shows | What pulls it down |
|---|---|---|---|
| Accuracy | How completely the repeated sentence matches the target words, order, and grammatical endings. | Nearly every target word is reproduced in the original order, with little or no added language. | Missing content words, substituted function words, reordered phrases, and dropped endings lower the score quickly. |
| Intelligibility | How clearly the repeat comes out — clear enough that what you said can be recognised word for word. | The transcript is easy to recover because the speech is clear, stable, and natural enough for reliable recognition. | Mumbled segments, clipped words, and long breakdowns reduce intelligibility even when the speaker remembers the right content. |
Each item comes back on a 0 to 5 scale, with these descriptions:
| Score | Label | What it means |
|---|---|---|
| 5 / 5 | Outstanding control | Complete, clear, and easy to follow with only trivial slips that do not reduce meaning. |
| 4 / 5 | Good control | Mostly complete and understandable, with minor omissions, hesitations, or language slips. |
| 3 / 5 | Developing control | Core meaning is present, but noticeable gaps in detail, delivery, or control reduce precision. |
| 2 / 5 | Limited control | Only part of the message lands clearly; frequent breakdowns force the listener to infer missing meaning. |
| 1 / 5 | Minimal control | Barely understandable or barely responsive, with severe limits in content or delivery. |
| 0 / 5 | No scorable response | Blank, off-task, copied, non-English, or too limited to score meaningfully. |
What ETS states
For TOEFL tests taken on or after January 21, 2026, official score reports use a 1–6 scale in half-band increments.
Sourced from the ETS 2026 test specifications, read (link confirmed ). Everything outside this box is Reach120’s own reading of it.
Where the marks actually go
Because the comparison is word against word, the losses are specific and repeatable. Almost all of them are one of four things.
| What goes wrong | Why it happens | What to do instead |
|---|---|---|
| A dropped function word | "Should", "the", "to" and "a" are unstressed in natural speech, so they are heard but not stored. | Rehearse the sentence in your head once before speaking. The function words survive the rehearsal; they do not survive a cold start. |
| A clipped ending | Rushing the last two words to finish inside the window, which cuts plurals and past-tense endings. | Slow the whole sentence rather than the start. A calm, complete repeat scores above a fast, truncated one. |
| A lost clause | Two-clause sentences overflow what you can hold, and the second half is what goes. | Hold the shape before the words: who did what, then the condition attached to it. |
| A substituted near-word | "Economy" for "economics", "assign" for "assigned" — plausible, and the comparison catches it. | Say the content words at full value even if it costs pace. Precision on the nouns is worth more than speed. |
Starting on the first word you caught
Weaker
The audio ends and you begin immediately, on whichever word arrived first, then reconstruct the rest as you go. The sentence comes out in the right order and two words short.
Stronger
You let a beat pass, recall the whole shape, and then speak once at an even pace. It feels slower and it scores higher, because what is scored is what you said, not how fast you began saying it.
This is the single most common habit among people who already speak English well. Fluency is what makes reconstruction feel safe, and reconstruction is what loses the function words.
Sample sentences, with what a correct repeat says
Sample sentences, with what a correct repeat says
Six sentences from the practice bank, printed with the repeat that counts as correct — which for this task is the sentence itself — and a note on the word most often dropped. These are not the sentences the drill page publishes, so you can work through both sets.
6 words
Remember to charge your recorder tonight.
What a correct repeat says
Remember to charge your recorder tonight.
Two content words carry the sentence — "charge" and "recorder". Say them at full value; the risk on a short sentence is rushing it and clipping the last word.
6 words
We should confirm the room number.
What a correct repeat says
We should confirm the room number.
"Should" is unstressed in natural speech and is the word most often dropped. Dropping it changes an obligation into a statement, and the transcript no longer matches.
10 words
The writing center closes early when the campus hosts events.
What a correct repeat says
The writing center closes early when the campus hosts events.
The subordinate clause — "when the campus hosts events" — is where the sentence is lost. Hold the whole shape before you start speaking rather than beginning on the first word you caught.
9 words
His presentation example made the economics concept much clearer.
What a correct repeat says
His presentation example made the economics concept much clearer.
Three long words in a row. Keep "economics" intact; substituting "economy" is a single-syllable slip that the comparison catches.
14 words
The biology lecture was canceled, so the lab instructor assigned a video reflection instead.
What a correct repeat says
The biology lecture was canceled, so the lab instructor assigned a video reflection instead.
Two clauses joined by "so". Repeat the connector — leaving it out produces two grammatical sentences that are not the sentence you heard.
13 words
After orientation, the international students met their tutors and toured the language center.
What a correct repeat says
After orientation, the international students met their tutors and toured the language center.
The sentence front-loads a time phrase, then runs two verbs off one subject. Miss "and toured the language center" and you have lost a third of the content.
These are Reach120 practice items written for the 2026 task format. None of them is a real TOEFL test item, and nothing on this page is published, licensed, endorsed or reviewed by ETS. The reasoning under each answer is ours.
How to practice it so the score moves
- 1Read a sentence above once, cover it, and say it from memory. That is the same demand the task makes, and it needs no microphone and no account.
- 2Check yourself word by word rather than by feel. "Close enough" is exactly the judgement the comparison does not make.
- 3When you drop something, write down WHICH word. After ten sentences the pattern is usually one category from the table above, not ten unrelated slips.
- 4Move up a tier only when the shorter sentences come back complete. Practicing above your holding capacity trains reconstruction, which is the habit you are trying to lose.
- 5Then take it to the drill, where your recording is transcribed and compared for you.
Listen and Repeat: common questions
- Does my accent matter on Listen and Repeat?
- What is scored is whether what you said can be recognised word for word against the sentence that was played. That is a question about clarity, not about which variety of English you speak.
- How long do I get to speak?
- ETS does not publish a per-item time limit for this task. Reach120 sets its own practice window, and that figure is our setting rather than an ETS rule.
- Is it better to say part of the sentence perfectly or all of it roughly?
- Both dimensions in the scoring table above carry weight — how much of the sentence came back, and how clearly it came out. A complete repeat at a calm pace serves both; a fast partial one serves neither.
- Can I practice this without a microphone?
- Yes, and it is worth doing. The sentences on this page are printed in full precisely so you can cover them and say them back on paper, on a phone, or on a train. The drill adds the transcription and the score.