How TOEFL Speaking Works in 2026

Eight minutes, eleven items, two parts

The Speaking section is the shortest part of the TOEFL. It runs eight minutes and contains eleven items. That is less time than you will spend on a single Reading module, and it is the reason so many candidates walk out of the test unsure what just happened. There is no warm-up, no long passage to settle into, and no place to hide a slow start. Eight minutes after the section opens, your entire Speaking band has been decided.

The eleven items split into two blocks that always run in the same order. The first block is Listen and Repeat: seven items where a sentence plays through your headphones and you say it back. The second block is Take an Interview: four questions asked by a pre-recorded interviewer, each answered into a 44 second recording window. There is no third block, no surprise task, and no optional section. If you know these two blocks cold, you know the entire section.

What makes this format hard is not difficulty of content. Nothing in Speaking requires academic knowledge, a specialised vocabulary, or an opinion you have to defend against a real person. What makes it hard is the pace. Every item starts on a timer you do not control, and every item ends on a timer you do not control. Your job across these eight minutes is to be ready to make sound the instant the machine asks for it, and to keep making useful sound until it stops.

This lesson maps the section end to end so nothing on test day is a surprise. The rest of the course then takes each block apart: how the scoring works, what a good repetition sounds like, how to shape 44 seconds of speech with no time to plan it, and which habits quietly hold most candidates a full band below where they should be.

The full item map

Every item you will meet, in the order it appears.

Part one: Listen and Repeat, items 1 to 7

Listen and Repeat is the quietest scoring opportunity on the test and the one candidates most often waste. A picture of a familiar place appears on screen: a university library, an advising office, a chemistry lab, a cafeteria, a train station. In our practice tests each of the sixteen Speaking sets uses one scene like this, and the seven sentences all belong to that world. Then a sentence plays once and the recording window opens. You say the sentence back.

The sentences start short and get longer. Early items are things like a one clause greeting. Later items carry two clauses and a subordinating word, which is where memory starts to fail people. The recording windows grow with the sentences, so you are never asked to say a twelve word sentence inside an eight second window, but you are also not given room to try twice.

Two things are being measured here and only two. First, did you reproduce the words accurately. Second, did you say them the way a fluent speaker would say them, with the right stress, rhythm and intonation. Nothing about opinion, content or creativity applies. This is the purest test of your ear and your mouth working together that the exam contains, which also makes it the fastest part of the section to improve. A candidate who is weak on the interview can still bank most of the available marks on these seven items inside a few weeks of daily practice.

The scoring language in our practice sets makes the ladder explicit. A perfect repetition with natural intonation and rhythm sits at the top. One minor error in pronunciation or rhythm sits one step down. Several errors with the sentence still intelligible sits in the middle. Significant errors that make you hard to understand sit near the bottom, and an unintelligible or missing response scores lowest. Notice that the gap between the top two levels is not about words at all. It is about music.

Part two: Take an Interview, items 8 to 11

The second block reframes the test as a conversation. A recorded interviewer introduces herself, explains that you have agreed to take part in a short research study, and then asks four questions about an everyday topic. In our practice sets the topics are deliberately ordinary: teamwork, outdoor activities, learning styles, leisure time, shopping habits, food, technology use, study habits, travel and so on. You are never asked to know something. You are asked to have a life and describe part of it clearly.

The four questions climb in difficulty in a fixed shape. Item 8 asks you to recall something personal, which is the most concrete and the easiest to fill with detail. Item 9 asks which of two things you prefer and why, which forces a clear choice. Item 10 puts a general claim in front of you and asks whether you agree, which needs a position and a reason. Item 11 is the broadest, asking about consequences, trade-offs or what should happen, which is where abstraction and time pressure meet.

Each answer is recorded for 44 seconds. The recording begins the moment the question audio finishes. You cannot pause it, you cannot restart it, and you cannot stop early to save time. When the 44 seconds are up the test moves on by itself. Four questions, four windows, no gaps in between that you control.

This is the part of the section that decides most bands, because it is where fluency, grammar, vocabulary and development are all visible at once. It is also where a small amount of structural training pays out enormously, because the difference between a candidate who fills 42 of the 44 seconds with organised content and one who stops at 22 seconds is usually not language ability. It is having a shape ready.

What the eight minutes feel like, in order

1. 1

   Setup and microphone check

   Before any scored item you confirm your microphone is working. Speak at the volume you actually intend to use in the test, not a cautious whisper. A check passed at whisper volume is a check that proves nothing, and a recording made at whisper volume is a recording the raters have to strain through.

2. 2

   Items 1 to 7, one sentence at a time

   The scene appears, a sentence plays, the recording window opens, you repeat. Then the next sentence. The rhythm is fast and mechanical. Do not carry a mistake from item 3 into item 4: each item is scored on its own, and dwelling on a fluffed word is how candidates turn one lost point into three.

3. 3

   The interview introduction

   The interviewer appears on video and sets up the research study framing. This is the only moment in the section where nothing is being recorded and nothing is being scored. Use it to sit up, unclench your jaw and take one slow breath. It is the last quiet moment you get.

4. 4

   Items 8 to 11, back to back

   Question audio plays, recording starts, you speak for 44 seconds, the test advances. Then the next question. There is no break where you can recover from a bad answer, so the recovery has to happen inside the first two seconds of the next one.

5. 5

   The section ends by itself

   After item 11 the section closes. There is no review screen, no chance to re-record and no confirmation that any individual answer was good. This is normal and it is not a sign that something went wrong.

Where the raw points sit

Speaking is built from 55 raw points: five points available on each of the seven Listen and Repeat items, which is 35, plus five on each of the four interview answers, which is 20. Those raw points are then converted to your reported band on the 1.0 to 6.0 scale.

Read that split again, because it changes how you should spend your practice time. Almost two thirds of the available raw points sit in the block that most candidates treat as a warm-up. Listen and Repeat is not the appetiser before the real test. It is the larger half of your Speaking score, and it is mechanically simpler to improve than interview fluency because there is exactly one right answer per item and you can hear immediately whether you produced it.

That does not make the interview optional. The four interview answers are where the ceiling is: a candidate who repeats sentences well but freezes for 44 seconds four times in a row cannot reach the top bands. But the fastest early gains almost always come from the seven repetition items, and a student with three weeks left should be spending real time there rather than only rehearsing opinions.

What you should be able to say yes to before test day

• I can describe both blocks of the Speaking section from memory, in order, with the item counts.
• I know that the interview recording starts the instant the question audio ends, with zero preparation time.
• I have practised repeating a twelve to fifteen word sentence after hearing it once, without writing it down.
• I have recorded at least one full 44 second answer and listened back to the whole thing without skipping.
• I have a first sentence I can produce for any interview question within one second of the audio ending.
• I have tested my actual microphone and headphones on the device I will use, at the volume I will use.
• I know that the section ends automatically and that there is no review screen, so I will not be waiting for one.

What the rest of this course covers

The next lesson opens the scoring up: what raters are actually listening for on each block, why an answer can be grammatically clean and still score in the middle, and what separates a four from a five on the same content. After that the course splits by task. Two lessons take Listen and Repeat apart, first the core technique of hearing and holding a sentence, then the stress, rhythm and intonation that decide the top of the scale. Two more take the interview apart, first the shape of the four question progression and then a framework you can run inside 44 seconds without preparation.

The final block is habits. Fluency and pausing, which is mostly about what you do instead of filler. Paraphrasing under time pressure, for the moment the exact word will not come. Recovery moves for the answers where your mind simply empties. Microphone and room setup, which sounds trivial and quietly costs real points. And a closing lesson on the habits that cap otherwise strong candidates at a band below their level.

Work through them in order if you have time. If you have a week, do this lesson, the scoring lesson, the interview framework lesson and the recovery lesson, and spend every spare fifteen minutes recording repetitions. That sequence gets the most points back for the least time.