How do I prepare for the four TOEFL Listening tasks?

What do you listen to in each of the four TOEFL Listening tasks?

In the four tasks you listen to a single short utterance, a short conversation between two people, a school announcement, and a short lecture, then answer multiple-choice questions. The task names are Listen and Choose a Response, Listen to a Conversation, Listen to an Announcement, and Listen to an Academic Talk (Test Overview, pages 4 and 8). Listening comes after Reading in the test, followed by Writing and Speaking (Test Overview, page 4).

Sources: what you hear and how questions work come from pages 8, 9, 11 and 12 of the Test Overview; the target question counts and CEFR (Common European Framework of Reference for Languages, six levels from A1 to C2) ranges come from the table on page 5 of the Specifications Document. The lecture length of 175 to 250 words is on page 20 of the ETS "TOEFL iBT Technical Manual" (hereafter "the Technical Manual", ETS Research Report RR-25-12, October 2025, checked 2026-10-30). The target question counts in the Specifications Document include unscored questions (Specifications Document, page 3, note 1).

Page 12 of the Specifications Document describes the listening screen this way. Each audio comes with a picture of the speaker, the recording is played once, and you answer the questions after the audio ends. Questions usually appear as a written question with multiple-choice options. The same page says the listening audio is produced with AI-generated voices and sometimes with human voice actors, and that it mixes US and Canadian, Australian, and British English accents and male and female voices in a roughly balanced way. Page 52 of the Technical Manual cites a study (Choi & Zu, 2022) finding that groups of test takers, such as by gender and age, responded similarly to AI voices and human voices on the same task.

You can take notes. The ETS "TOEFL iBT Listening Section" page (ets.org Listening page, no publication date, checked 2026-10-30) says you may take notes while listening and use them to answer the questions. The FAQ in ETS "TOEFL iBT Test Content" says test centers provide paper and pencils, while the at-home Home Edition uses a small whiteboard, or one blank sheet of paper inside a transparent plastic sheet protector, with an erasable marker (no publication date, checked 2026-10-30). Page 10 of the 2026-27 "TOEFL iBT Information Bulletin" (no publication date, in effect July 2026 to June 2027) says the test center paper cannot be used before the test, between sections, or during breaks, and says there is no scheduled break.

One task name is easy to confuse. Listen and Repeat has Listen in its name, but it is a Speaking task and is not one of the four Listening tasks (Test Overview, page 4). The full structure of all four sections is in the Unlock Prep (unlockprep.com) guide The 2026 TOEFL iBT four sections: what ETS has and has not disclosed.

How many questions from each task count toward the Listening score?

Each Listening question earns 1 point if correct and 0 if wrong, and every path has 35 scored questions, but how they split across tasks depends on the second module. The 1 point and 0 point scoring is on page 34 of the Technical Manual and page 13 of the Specifications Document, and the 35 questions and per-task split are in Table 1 on page 29 of the Technical Manual. The Specifications Document describes Listening as machine scored (scored by fixed rules that compare answers with the key) (page 2 table, page 3 note 4).

Listening is two-stage adaptive (the difficulty of the second module depends on your answers in the first module). Page 28 of the Technical Manual says the first module (the router) is built from tasks of medium difficulty (CEFR B1 or B2), and that the difficulty of the second module is set by performance on the first. Page 3 of the Test Overview gives the purpose as matching task difficulty to the test taker's ability level. Each test taker takes one of two paths: "first module + lower module" or "first module + upper module" (Technical Manual, page 28).

Source: Technical Manual, page 29, Table 1 (scored question counts, checked 2026-10-30). For time, the tables on pages 2 and 5 of the Specifications Document give 18 minutes for the first module, 7 minutes for the lower module, and 11 minutes for the upper module. Added together, that is 25 or 29 minutes (our calculation).

Unscored questions are mixed in as well. Page 28 of the Technical Manual says Listening modules can contain 12 unscored questions, and it does not say how many are in each module or which questions they are. Adding 12 to the 35 scored questions matches the 47 questions on page 4 of the Test Overview, but that sum is our own reconciliation.

The number of questions you get right does not become your score directly. Page 34 of the Technical Manual says the Listening raw score (the sum of points on correct questions) is converted to a scaled score through IRT (item response theory) equating (a statistical procedure that puts scores from test forms of different difficulty on the same scale), that this process reflects the difference in difficulty of the second module, and that the reported score is not the same as the number or percentage correct. On the score report, Listening appears on the 1 to 6 band in steps of 0.5 (Test Overview, page 21). The raw score range differs between documents. The table on page 2 of the Specifications Document and Table 8 on page 41 of the Technical Manual give 35 (0 to 35), while Table 1 on page 21 of the Test Overview gives 0 to 30. How getting the upper module relates to your score is covered in TOEFL adaptive testing: can your score be low after an upper module?. When the test ends, unofficial Reading and Listening scores appear on screen, and they may differ from the official scores because of statistical analysis after the test (Information Bulletin, page 11).

We read this table in two ways. On the harder path, 12 of the 35 questions are Academic Talk questions, and on the easier path, Listen and Choose a Response and Announcement together make up 23 (question counts from Table 1; the sums and interpretation are ours). Which path you get is decided by your first-module result, and the first module contains all four tasks, so in our view practicing only one task leaves gaps on either path.

Where do you lose points in each task?

The places where we see points lost each attach to one line of the abilities ETS lists for that task. In our view, questions are easy to lose on the structure and implied meaning of an utterance in Listen and Choose a Response, on connecting the speaking turns in Conversation, on separating changed information from unchanged information in Announcement, and on reading the role an example plays in Academic Talk. The scripts, questions and options below are examples the Unlock Prep editorial team wrote for illustration and do not reproduce ETS questions. The ability wording in the "Why" lines is our own paraphrase of the lists on pages 8, 9, 11 and 12 of the Test Overview, and the answers are ours.

Example 1: Listen and Choose a Response

• What you hear (not shown as text on screen): Weren't you supposed to turn in your lab report today?
• Options: A. Yes, the lab is on the second floor. / B. The professor gave us until Friday. / C. Sure, I can turn it in for you. / D. No, I wasn't in the lab today.
• Wrong option: A. It starts with Yes and has the shape of an answer to the question, but its content is the location of the lab.
• Right option: B. Without a yes or no, it answers "weren't you supposed to turn it in today?" with the fact that the deadline was moved to Friday.
• Why: Page 8 of the Test Overview says this task measures simple grammatical structures, including question forms, socially appropriate responses, and the implied meaning in what is said. The negative question (Weren't you …) asks whether the report was submitted, and B is the only response that fits that question. D only answers about the lab and says nothing about the report.

Example 2: Listen to a Conversation

• Script: (Woman) Are you going to the career fair this afternoon? (Man) I was planning to, but my résumé still lists the café job I quit in May. (Woman) The writing center reviews résumés without an appointment until two o'clock. (Man) Oh, then I'll print a copy and stop by there first.
• Question: What will the man probably do next?
• Options: A. Go straight to the career fair / B. Take his résumé to the writing center / C. Apply for a job at a café / D. Make an appointment for tomorrow
• Wrong option: A. This is the original plan from the man's first turn.
• Right option: B. In his last turn, the man changes his plan: he will print his résumé and stop by the writing center first.
• Why: Page 9 of the Test Overview says this task measures the ability to make simple predictions about what a speaker will do next and to follow the connection of ideas across speaking turns. If you catch only the first turn and let the last one slip, you pick A. D contradicts "without an appointment" in what the woman said.

Example 3: Listen to an Announcement (an announcement we wrote, 56 words)

• Script: Good morning, everyone in Biology 210. Tomorrow's lab won't meet in Room 118. A ventilation repair is scheduled there, so we'll use the teaching lab on the ground floor of Carver Hall instead. We'll still start at nine. One more thing: the spare safety goggles are locked in Room 118, so please bring your own pair.
• Question: What does the speaker imply about tomorrow's lab?
• Options: A. It will start later than usual. / B. Students who forget goggles will not be able to borrow a pair. / C. It will be held in Room 118 after the repair. / D. It has been canceled because of the repair.
• Wrong option: A. The location changed, but the start time is still nine.
• Right option: B. The spare goggles are locked in Room 118 and the lab is in another building, so there is nowhere to borrow them. The announcement never says this conclusion out loud.
• Why: Page 11 of the Test Overview says this task measures the ability to understand important details in a short announcement and to infer meaning from information that is not stated directly. If you remember what changed (the location) and what stayed the same (the time) as one lump, A sounds plausible.

Example 4: Listen to an Academic Talk (a lecture we wrote, 177 words)

• Script: Walk across almost any park and you'll notice narrow dirt trails cutting across the grass. Planners call these desire paths. They form when enough people decide the paved route is too long and take a shortcut instead. For years, many cities treated desire paths as damage. They put up fences or planted bushes to block them, and then paid crews to repair the fences and replant the grass when people walked around them anyway. But some planners now see these trails as free information. A desire path shows exactly where people want to walk, something a survey of walkers might miss. Imagine a new campus that opens without any paved walkways at all. After a year, the grounds staff can simply pave the routes that students have worn into the lawn. Of course, not every shortcut should become a sidewalk. Some paths cross ground that floods easily, and others damage tree roots. So the point is to treat each trail as a question. Why are people choosing this route, and what is the official path getting wrong?
• Gist: The lecture moves from seeing the shortcuts people wear into lawns (desire paths) as damage to seeing them as information that shows where people actually want to walk.
• Question: Why does the speaker mention a campus that opens without paved walkways?
• Options: A. To show that students damage lawns more than other people do / B. To give an example of using desire paths to decide where to build walkways / C. To argue that every shortcut should be paved / D. To explain why surveys of walkers are expensive
• Wrong option: C. In the very next sentence (Of course …), the speaker rules out the idea of paving every shortcut.
• Right option: B. The campus example comes right after the claim that shortcuts are "free information" and supports that claim.
• Why: Page 12 of the Test Overview says this task measures the ability to understand the main ideas and supporting ideas of a lecture and its organizational features. If you remember only the example and let the claims and signals around it (But, Of course, So) slip past, you read the example as the lecture's conclusion. In the real test, each lecture has 4 questions (Test Overview, page 12).

The wrong options in all four examples contain words from the script, so memory of what you heard is not enough to filter them out. In our view, what separates the right answer is whether those words attach to the question, a speaking turn, or a claim.

What has ETS not disclosed, and what differs between documents?

ETS has disclosed the task formats, target question counts, 1 point and 0 point scoring, scored question counts by path, single playback of recordings, and that note-taking is allowed, while the time limit per question and whether you can go back to earlier questions are not in the ETS documents we checked.

We did not find the following in pages 1 to 28 of the Test Overview, pages 1 to 18 of the Specifications Document, pages 16 to 21, 28 to 29, 34, 41, 43 and 52 of the Technical Manual plus a full-text word search, the 2026-27 Information Bulletin, or the ets.org Test Content and Listening pages (checked 2026-10-30).

• A time limit for each question or each task (there are only expected times per module)
• Whether you can go back to questions you have answered and change them
• The number of questions attached to each announcement in Listen to an Announcement
• How many first-module questions you must get right to reach the upper module (the cut score)
• The numbers in the conversion table from raw scores to the 1 to 6 band (page 34 of the Technical Manual describes only the method)
• Which questions are unscored
• Which audio uses AI voices and which uses human voice actors
• Whether questions are shown before you hear the audio (page 12 of the Specifications Document says only that you answer after the audio)

Here is where the ETS documents differ from one another.

Time is also stated differently in each place. Page 4 of the Test Overview and the Test Content table give about 29 minutes, while the Specifications Document says to add 7 or 11 minutes for the second module to 18 minutes for the first module (page 2 table, page 3 note 5). The sentence below the Test Content table says time and questions can vary as the test adapts. Section times are collected in How long is the 2026 TOEFL iBT?. The Specifications Document still contains a line saying minor revisions may be made before launch (page 3). We did not cross-check articles from other companies or blogs.

How do I practice the four tasks?

In our view, the order that fits the format of this test is to play each recording only once, answer with a single sheet of notes, and then write down one line of the measured abilities for each question you missed.

1. Do not replay practice recordings. This is because page 12 of the Specifications Document says each recording is played once. Listen again only after you have finished scoring.
2. Solve Listen and Choose a Response by ear only, without looking at a script. Before reading the options, first decide what kind of utterance you heard (yes or no question, wh- question, negative question, request, or suggestion). Ruling out A in Example 1 is this step.
3. For Conversation and Announcement, split your notes into two columns, one for each speaker, or two columns labeled "changed / unchanged". The plan change in Example 2 and the location change in Example 3 go in these columns. Page 12 of the Specifications Document says you answer the questions after the audio and does not say whether the questions are shown before you listen. So practice listening without knowing what will be asked.
4. For Academic Talk, when an example comes up, write one line on "what is it meant to show", and mark where direction-changing words such as But, Of course and So appear. The "organizational features" line on page 12 of the Test Overview is what this practice targets.
5. Match your note-taking tools to what you will use on test day. Test centers use paper and pencil, and the Home Edition uses a small whiteboard or one sheet of paper inside a plastic sheet protector with an erasable marker (Test Content FAQ, Information Bulletin page 10). A single sheet in a sheet protector leaves little room to write, so practice noting things one word or one arrow at a time.
6. After each practice set, record your wrong answers by task, as in the table under "How many questions from each task count toward the Listening score?" above. In our view, many Academic Talk errors cost you more on the harder path, and many Listen and Choose a Response or Announcement errors cost you more on the easier path.