The shape of the new speaking section
Since the January 2026 update, the speaking section is 11 items in about 8 minutes, of two kinds:
| Task | Items | What you do |
|---|---|---|
| Listen and Repeat | 7 | You hear a sentence and say it back. |
| Take an Interview | 4 | You answer questions in an interview scenario. |
Each item is worth a maximum of 5 points, and the section is reported on the 1–6 band scale along with the other three. If you are preparing from a guide that describes six speaking tasks and a 0–30 score, it is describing the previous test.
Unlike reading and listening, the speaking section is linear — everybody taking a given form gets the same tasks, and your performance on the early items does not change what comes next.
The word in the rubric that decides your band
ETS's blueprint describes the skill this task measures as responding to questions with clear, coherent elaboration, using accurate grammar, varied vocabulary and intelligible prosody.
Elaboration is the load-bearing word. The question is a prompt to speak, not a question to be disposed of — and this has a consequence that surprises people:
The CEFR descriptors ETS publishes for this task make the ladder explicit. At B1, a candidate responds with some elaboration and maintains a conversational pace with occasional hesitation. At C1, they respond fluently and spontaneously to complex questions, expressing nuanced ideas. The distance between those two is almost entirely how much is said after the answer arrives.
How to elaborate without padding
Elaboration is not talking for longer. Repeating yourself in different words is padding, and it is visible. What the descriptors are asking for is a second layer — and there are only a few kinds, so you can have them ready.
| After you answer, add… | Sounds like |
|---|---|
| A reason | …mostly because I get distracted at home — the kitchen is right there. |
| A specific instance | …last term I tried working in my room for a week and got almost nothing done. |
| A contrast | …although if I'm just reading, home is fine; it's writing that needs the library. |
| A condition | …it depends on what I'm working on, really. |
| A consequence | …so I've ended up going in every day, which has changed my whole routine. |
Two of those, attached to a direct answer, produce a response that is fluent, developed, and about the right length — without a memorised frame.
And notice what they do to the grammar. Although…, it depends on whether…, so I've ended up… are subordinate clauses, a conditional and a perfect tense. The elaboration is what produces the range; you do not have to insert complex structures deliberately, you have to say more about the thing you just said.
The questions get harder as you go
The blueprint is explicit that difficulty increases across the task: prompts move from brief factual statements toward opinion, explanation, prediction and narration.
So the four interview questions are a ramp, not four samples of the same thing. Something like:
- Factual — How do you usually get to campus?
- Opinion — Do you think a campus should be car-free?
- Explanation / prediction — What might happen if the university closed the car park?
- Narration — Tell me about a time a change like that affected you.
The hardest of the four is usually the narration. It is worth practising specifically, because a story has a structure that a two-sentence opinion does not — and because narrating pulls past tenses out of you automatically, which is exactly the range being looked for.
Prosody is in the rubric — and it is not your accent
The skill statement names intelligible prosody. That is unusual: most test rubrics say "pronunciation" and leave the reader to guess. Prosody is a narrower and much more useful word.
It means the music of speech — stress, rhythm and intonation — as opposed to the individual sounds. And it is the part of pronunciation that carries meaning:
- Sentence stress. I didn't say she took the money means six different things depending on which word you lean on.
- Intonation. A falling tone finishes a thought; a rising one leaves it open. Ending every sentence on a rise makes a confident answer sound uncertain.
- Rhythm. English compresses its unstressed words (a, of, to, and). Giving every word equal weight is what makes speech sound effortful, even when every sound is correct.
None of that is about having a particular accent. The descriptors ask whether your speech is intelligible and whether stress and intonation convey meaning. A strong regional accent with good prosody is easy to listen to. Flat, evenly-stressed speech with textbook vowels is not.
It is also the most improvable part of pronunciation, because it does not require you to relearn a single sound — only to lean on the words that matter.
And the other seven items: Listen and Repeat
Seven of the eleven speaking items simply ask you to say back a sentence you have heard. The sentences start short and simple and grow: compound verbs, dependent clauses, relative clauses, complex noun phrases.
It looks trivial and it is not, because you cannot repeat what you did not parse. Holding a fifteen-word sentence with two clauses in memory long enough to say it back requires that you understood its structure — which is precisely why it is a valid measure, and why it gets harder rather than longer.
- Listen to the whole sentence before you begin. Starting to speak while it is still playing costs you the end of it.
- Reproduce the rhythm, not just the words. The task is scored on intelligibility and prosody — a flat, word-by-word recitation of the right words is not a good repetition.
- If you lose a word, keep the sentence going. A fluent near-miss beats a breakdown, and stopping dead scores nothing at all.
Frequently asked questions
What is TOEFL Take an Interview?
It is one of the two speaking tasks on the TOEFL as updated in January 2026. You answer four questions in an interview scenario, which together with seven Listen and Repeat items makes up an 11-item speaking section lasting about 8 minutes. Each item is worth a maximum of 5 points, and the section is reported on the 1 to 6 band scale.
How long should my Take an Interview answers be?
Long enough to elaborate. ETS describes the skill as responding with clear, coherent elaboration — so a correct but short answer, however grammatical, gives the scorer very little to score. Answer the question directly, then add two of the following: a reason, a specific instance, a contrast, a condition, or a consequence. That is usually the right length, and it produces complex grammar on its own.
Do the TOEFL interview questions get harder?
Yes. The official blueprint says difficulty increases across the task, with prompts moving from brief factual statements toward opinion, explanation, prediction and narration. The common mistake is to spend your energy elaborating on the easy first question and arrive at the narration item with nothing left — but the later items are exactly where range and nuance are meant to be demonstrated.
What is prosody, and why does TOEFL score it?
Prosody is the music of speech — stress, rhythm and intonation — as opposed to the individual sounds, and ETS's skill statement for this task explicitly names intelligible prosody. It carries meaning: a falling tone finishes a thought while a rising one leaves it open, and sentence stress alone can change what a sentence means. It is not about your accent, and it is the most improvable part of pronunciation because it requires relearning no sounds at all.
Is the TOEFL speaking section adaptive?
No. Reading and listening are two-stage adaptive tests, but speaking and writing are linear — everyone taking a given form receives the same tasks, and your performance on early items does not change what comes next.