The new TOEFL Listening section: four task types, how it adapts, with sample questions
TOEFL Listening since January 2026: the four task types, played once with no replay, the two adaptive parts, and a sample of each with answers.
Updated 6 November 2026 · 10 min read
The short answer
TOEFL Listening since January 2026 has four task types: Listen and Choose a Response, one spoken line and four written replies; Listen to a Conversation, two speakers and two questions; Listen to an Announcement, one speaker and two questions; and Listen to an Academic Talk, a short lecture with four questions. Up to 47 questions in about 29 minutes, every recording played once, no replay and no going back, in two parts where the second depends on the first.
What changed
The old Listening section was three or four lectures of five or six minutes and two or three conversations, with six and five questions each, note-taking allowed, 36 to 41 minutes. The 2026 section is shorter and built from more, briefer recordings: a one-line prompt you answer with a reply, a 60-word exchange, a 60-word announcement, a 200-word talk. Nothing lasts more than about a minute and a half.
| Before 2026 | Since January 2026 | |
|---|---|---|
| Recordings | 3–4 lectures (5–6 min) and 2–3 conversations (3 min) | One-line prompts, conversations of 50–90 words, announcements of 40–70 words, talks of 175–250 words |
| Questions | 28–39 | Up to 47, of which 35 count |
| Time | 36–41 minutes | About 29 minutes, in two timed parts |
| Replay | No | No; every recording plays once |
| Adaptive | No | Yes: the second part depends on the first |
| Score | 0 to 30 | Band 1 to 6 |
ETS's 2026 specification describes the section as listening across everyday and academic situations from CEFR A1 to C2, with speakers in North American, British and Australian accents at a natural pace.
How the two parts work
The first part is of moderate difficulty and lasts about 18 minutes: eight Choose a Response items, two conversations, two announcements and one talk. Your result decides the second part. The lower second part, about 7 minutes, has seven more Choose a Response items, two conversations and two announcements, and no talk. The upper second part, about 11 minutes, has three Choose a Response items, two conversations and two talks, and no announcements. The scoring model knows which second part you took, so the band reflects your level whichever way you were routed. The adaptive Reading and Listening guide explains the routing in more detail.
Three rules shape how you work: each recording plays once, with no player to pause or replay; each question has its own clock; and there is no Back button. The speaker's picture is on screen while the audio plays, and the question appears afterwards with its printed stem. You cannot see the question before the recording, so you listen for everything, not for one answer.
Listen and Choose a Response
How it works
You hear one person say one line, a question or a statement of 4 to 12 words, and choose the best of four written replies. The line is never shown. The situations are everyday: work, home, errands, travel, plans. The best reply is often indirect: a counter-question, a correction that starts "Actually", an offer, an acceptance. The distractors are built from the ways people mishear: a reply that answers a different question word, a reply that uses the same word with a different meaning ("book" the verb against "book" the noun), a reply in the wrong form ("Yes, I would" to "Did you lock the door?"), and a reply that is on topic but not a reply at all.
A sample
You hear a woman say: "Weren't you supposed to be at the dentist this morning?"
- They moved it to Friday. ✓
- The dentist is on Main Street.
- I was supposed to, yes, I will.
- No, it doesn't hurt anymore.
The second option answers "where", the third is the question's own words in the wrong shape, the fourth replies to a question about pain that was never asked.
Listen to a Conversation
How it works
Two speakers, four to six turns, 50 to 90 words, on a campus or in daily life: a shop, a service desk, a plan for the weekend, a problem and a way round it. Two questions follow. One is nearly always about what a speaker will do next, set up by the last turn; the other asks about a problem, something a speaker did not know, or what a speaker means by a particular line whose literal meaning is not its point.
A sample
Listen to a conversation.
Man: Are you still selling your old bike? Woman: I was, but my sister's borrowing it this term. Man: Oh. I start a job across town on Monday, and the bus takes forever. Woman: The bike shop on Elm Street rents them by the month, actually. Man: Really? I'll go by after class.
- What is the man's problem?
- He cannot find his bike
- He needs a way to get to a new job ✓
- He missed the bus this morning
- He has to work on Monday
- What will the man most likely do next?
- Borrow the woman's bike
- Take the bus to class
- Visit a bike shop ✓
- Talk to the woman's sister
Listen to an Announcement
How it works
One speaker, 40 to 70 words, in an academic setting: a classroom, a club meeting, the campus radio, a school event. The content is practical: a schedule, a room change, a deadline, a contest, a guest lecture. Two questions: what the announcement is about, and either why the speaker mentions something or what the speaker wants listeners to do. Exact times and places in the recording are there to serve as distractors, so hold on to them.
A sample
Listen to an announcement in a classroom.
Good morning, everyone. Two quick things before we start. First, next Thursday's class will meet in the computer lab on the second floor, not here, because we'll be working with the mapping software. Second, the field trip sign-up sheet is on my desk. There are twelve places, and they go in the order people sign, so if you want to come, do it today.
- What is the main topic of the announcement?
- A change of room and a sign-up sheet ✓
- A new software course
- A problem with the computer lab
- The results of a field trip
- What does the speaker encourage the students to do?
- Bring a laptop to the next class
- Sign up for the trip today ✓
- Move to the second floor now
- Download the mapping software
Listen to an Academic Talk
How it works
One speaker, 175 to 250 words: a professor in class or a podcast host, on history, art, science, business or social science. No background knowledge is needed. The talk usually opens with a hook or a question, defines its term in plain words, gives two or three examples or a contrast, and ends by pointing to what comes next. Four questions: the main topic, a detail or point the speaker makes, why the speaker mentions something, what the speaker implies, or what the speaker will discuss next. The main-topic key is a plain paraphrase, not the technical name; the distractors are narrower or broader topics mentioned in passing.
A sample
Listen to a talk in a biology class.
So, how do we know how old a tree is without cutting it down? The usual answer is tree rings, uh, growth rings: each year a tree adds a layer of wood, lighter in spring when it grows fast, darker in late summer when it slows. Count the pairs and you have the age. But rings tell us more than age. A wide ring means a good year, plenty of rain; a narrow one means drought or cold. Put the rings of many trees together and you get a record of the climate going back centuries, long before anyone kept weather records. Scientists have used old beams in buildings and even timber from shipwrecks to extend that record. And here's the part I like: because the pattern of wide and narrow years is the same across a region, a piece of wood can be matched to the exact years it grew, like a fingerprint. Next time, we'll look at how that matching is done.
- What is the talk mainly about?
- How to measure the height of a tree
- What tree rings can tell us ✓
- Why some trees grow faster than others
- How ships were built from timber
- According to the speaker, what does a narrow ring indicate?
- A year with little rain or low temperatures ✓
- A young tree
- Wood from a shipwreck
- A tree that was cut down early
- Why does the speaker mention old beams in buildings?
- To explain how buildings were dated
- To illustrate how the climate record can be extended further back ✓
- To argue that old timber should be preserved
- To show that trees were once much larger
- What will the speaker most likely discuss next?
- How to count rings accurately
- How a piece of wood is matched to the years it grew ✓
- Why weather records began
- Which trees live longest
The second question shows a common trap: "wood from a shipwreck" is in the talk, but it answers a different question.
How the section becomes a band
One point per correct answer, 35 scored questions, and a conversion that accounts for which second part you took. The result is a Listening band from 1 to 6 in half steps, averaged with the other three sections for the overall band. On ETS's concordance a Listening 5 corresponds to 22–25 on the old 0–30 scale and a 4 to 17–19.
How to practise Listening
The habit that matters most is listening once, without a transcript, and answering from memory, because that is the only way the test works. Practise each task type on its own: short replies train you to hear the question word and the intonation; conversations train you to notice the last turn; talks train you to follow a shape instead of catching every word. Skorway's practice sets play every recording once, as the test does, with real voices and an explanation for every wrong answer, and the free mock runs the adaptive section against the per-question clock.
How many questions are in TOEFL Listening 2026?
Up to 47, of which 35 are scored; the rest are trial questions you cannot tell apart from the real ones.
How long is the TOEFL Listening section?
About 29 minutes: a first part of about 18 minutes and a second of 7 or 11, depending on which part you are routed to.
Can I replay a recording in TOEFL Listening?
No. Every recording plays once, and there is no pause. The question appears after the recording ends.
Can I take notes in TOEFL Listening?
ETS allows note-taking throughout the test, but the recordings are short enough that most candidates do better listening than writing; a 60-word conversation is over before a note is finished.
See your Listening band before test day
A transcript on a page is not the test. Skorway's free mock plays the full Listening section once, as on test day, with real voices, the per-question clock and no Back button, routes you to the second part the way ETS does, and scores it within minutes with an explanation for every question you missed. You find out which of the four task types costs you marks before the real test does. No credit card.
Checked against ETS's 2026 specification and technical manual on 6 November 2026. The sample recordings are Skorway's own, not ETS's. TOEFL® is a registered trademark of ETS. Skorway is not endorsed or approved by ETS.