TOEFL Speaking Practice for the New 2026 Format
Most of what comes back when a teacher searches for TOEFL speaking practice right now is describing a test that no longer exists. On January 21, 2026, ETS rebuilt the TOEFL iBT from the ground up — shorter, adaptive, and rescored on a new band system — and the Speaking section didn't just get trimmed, it got replaced outright. The old four-task structure (one independent opinion question, three integrated read-then-listen-then-speak tasks) is gone. Confirmed directly on ETS's own TOEFL iBT content page: the current Speaking section runs on exactly two task types, contains 11 items, and takes roughly eight minutes — down from around seventeen. Half of that new structure maps cleanly onto ordinary spontaneous-speaking classroom practice. The other half asks for something genuinely different, and pretending otherwise isn't useful to anyone actually prepping a class for the real test.
What actually changed, task by task
Here's the actual TOEFL Speaking format 2026 test-takers sit today, cross-checked across ETS's own site and four independent test-prep sources that all agree on the specifics.
Listen and Repeat comes first: seven prompts of increasing difficulty. A sentence plays, tied to a simple visual, and the candidate repeats it back exactly — no preparation time, 8 to 12 seconds to respond per item. ETS's own materials describe what it measures as pronunciation, accuracy, and memory: can the candidate hold a sentence in working memory long enough to reproduce it intelligibly under time pressure. It's closer to a controlled listening-and-recall drill than anything resembling a "speaking task" in the older sense.
Take an Interview follows: four follow-up questions on familiar topics, delivered conversationally, again with no preparation time — but 45 seconds to respond per item, roughly what the old Independent Speaking task gave. ETS frames this as measuring fluency, elaboration, and vocabulary: not a rehearsed answer, but the ability to keep developing an idea naturally when a real interviewer's next question could go anywhere.
Both tasks share one property the old format didn't have at all: zero prep time. The old test gave 15 to 30 seconds of silent planning before every response. The new one doesn't give any — candidates are expected to speak the instant a prompt ends, on both tasks. That's arguably the single biggest pedagogical shift buried in the redesign, and it's the one most existing "TOEFL speaking topics" content hasn't caught up to yet.
Why old prompts are still circulating
A B2 exam-prep student came in this spring with a printed set of practice questions pulled from a study guide — the classic "describe a person who has influenced you" Independent Speaking prompt, the kind that used to open every TOEFL Speaking section. We ran it through Spin & Speak as a 60-second monologue, which felt like solid prep at the time. It wasn't until checking the actual current test together, a few weeks later, that we realized ETS doesn't ask that question anymore — the entire independent-opinion-essay format it came from was retired in January. The study guide hadn't been updated, and neither, honestly, had my own assumption about what "TOEFL speaking practice" meant. Search "Take an Interview TOEFL" right now and most of what comes back still describes the old test, not the one students actually sit. That gap between what's searchable and what's real is exactly what this post is trying to close.
Take an Interview: the half that transfers to normal classroom practice
Ask, Tell, Reveal generates a speaking action — Ask, Tell, or Reveal — on a rotating theme, with a chain-reaction mechanic that fires a natural follow-up prompt based on whatever the student just said. That chain-reaction shape is a close, honest match for Take an Interview's real mechanic: an examiner-style question lands, the candidate answers immediately, and the next question responds to what they said rather than following a fixed script. The app's own theory grounding — Austin (1962) and Searle (1969)'s speech-act theory — treats Ask, Tell, and Reveal as three distinct illocutionary types requiring different discourse resources, which maps directly onto the range of familiar-topic follow-ups Take an Interview actually asks (home, study habits, daily routines, preferences), not one repeated question shape.
The zero-prep constraint is the part worth deliberately protecting when running this app for TOEFL prep specifically. Ask, Tell, Reveal's own convention is to nominate a student, read the prompt, and let them respond — there's no built-in pause, which is the right setting here. Resist the instinct to add a "thinking time" buffer before nominating, even for lower-level classes; TOEFL's real Take an Interview task gives none, and rehearsing with an artificial pause trains a habit the actual test will punish.
Building the "no notice" reflex
Two more apps round out the elaboration and immediacy side of Take an Interview prep without duplicating what Ask, Tell, Reveal already covers.
Spin & Speak gives students 5 seconds of prep and 60 seconds to speak on a topic without stopping, using keyword scaffolds as a safety net rather than a script. It's a longer window than Take an Interview's 45-second response, which is the point: a student who can sustain 60 seconds of unrehearsed output with 5 seconds' notice has real headroom left for a 45-second answer with none. Nation and Newton (2009), already cited in the app's own documentation, ground this in fluency-through-timed-pressure — exactly the automatic-retrieval condition Take an Interview is built to test.
Reaction Reactor, run in Keyword Catalyst mode, presents a scenario first and asks students to produce a response before the target expression is ever revealed. That ordering — respond first, see the model second — is a deliberate rehearsal of answering cold, without knowing in advance what "correct" looks like. It's a smaller, faster activity than the other two, best used as a five-minute warm-up immediately before an Ask, Tell, Reveal round, to put the class into a no-notice mindset before the main activity starts.
| App | Take an Interview skill rehearsed | CEFR range | Formation | What it trains |
|---|---|---|---|---|
| Ask, Tell, Reveal | Examiner-style follow-up questions on familiar topics | A1–C1 | Free Circle | Immediate, unscripted answers that respond to what was just said |
| Spin & Speak | Sustained elaboration under time pressure | A1–C1 | Whole Class | Extending an answer well past the first sentence |
| Reaction Reactor | Answering before seeing a model | A2+–C1 | Whole Class | The reflex of responding immediately, not waiting for a script |
A ten-minute weekly rotation
Open with a Reaction Reactor round in Keyword Catalyst mode — two or three scenarios, no scaffolding, straight to a student response before revealing the target expression. This is the warm-up, not the main event: its job is to put the room in a "respond now" mindset. Move into Ask, Tell, Reveal for the bulk of the session — six to eight rounds, using the chain-reaction button liberally so prompts build on real student answers instead of resetting to a fresh, unconnected topic each time. Close with one Spin & Speak round per student pair or small group, using the shortest prep setting available, so the 60-second monologue still feels closer to Take an Interview's zero-notice demand than to a rehearsed presentation. That sequence takes roughly ten minutes and rehearses the actual shape of the new task — immediate response, developing elaboration, familiar but unpredictable topics — rather than the debate-prep or essay-structure habits that TOEFL speaking topics used to reward under the old independent-task format.
What this doesn't replace, and how TOEFL now differs from Cambridge and IELTS
Be honest about the half this doesn't cover: nothing in ENK's library currently drills Listen and Repeat directly. A "Listen and Repeat TOEFL" item asks a student to hold an exact sentence in working memory for a few seconds and reproduce it verbatim, under a genuinely tight clock, with pronunciation and word-for-word accuracy both scored. That's an echoic-memory and articulation task, not a generative-speech task — a fundamentally different skill from anything the apps above are built to train, since ENK's speaking apps are all designed around producing new language, not reproducing a fixed sentence exactly. Seven of the section's eleven items are Listen and Repeat, so it's the majority of the test by item count even though it's the smaller half of what a normal ESL classroom already practices. A dedicated echo-and-repeat drill — reading a sentence aloud once, giving students eight to twelve seconds to reproduce it exactly, then moving on — is worth running as a separate five-minute warm-up alongside this rotation, even without app support for it yet.
It's also worth being explicit about how the new TOEFL compares to the other two major English exams already covered here. Cambridge Speaking Exam Practice covers a face-to-face, two-examiner format with a genuine collaborative task between paired candidates — something neither TOEFL nor IELTS has at all. IELTS Speaking Part 2 Practice covers a solo two-minute long turn with a full minute of silent note-taking beforehand — the opposite of TOEFL's zero-prep constraint on every item. TOEFL's new Speaking section, uniquely among the three, gives no preparation time anywhere and delivers both tasks through a computer rather than a live interlocutor. Don't run IELTS's long-turn prep or Cambridge's paired-task prep as TOEFL practice; the no-notice constraint is the one thing that makes this test's Speaking section genuinely different from the other two.
Every app here lives inside the standard ENK account, gated by the same monthly AI credits as the rest of the library rather than a separate exam-prep tier — 2,300 credits free every month, or from €5/month for 5,700 credits on the pricing page. Browse the full app library to see how these three sit alongside the rest of the speaking-focused apps covered in ESL Speaking Activities.
FAQ
What changed in the TOEFL Speaking section in 2026? Effective January 21, 2026, ETS replaced the old four-task Speaking section (one independent opinion task plus three integrated read-then-listen-then-speak tasks) with two new task types: Listen and Repeat and Take an Interview. The section shrank from around 17 minutes to about 8, and neither new task gives any preparation time.
Is TOEFL Speaking practice from before 2026 still useful? Only partly. The old Independent and Integrated Speaking prompt banks no longer match what the test actually asks — the reading-and-listening-then-speaking format is gone entirely. Classroom fluency and elaboration practice still transfers to Take an Interview, but rehearsing old-format questions specifically is now training for a task that doesn't exist.
What does Listen and Repeat actually test? Seven sentence-repetition prompts of increasing difficulty, each with no prep time and 8 to 12 seconds to respond. ETS describes it as measuring pronunciation, accuracy, and memory — the ability to hold a sentence and reproduce it exactly under time pressure, not to generate new language.
Do these apps replace dedicated TOEFL prep materials? No. They build the underlying fluency and no-notice response skill that Take an Interview specifically tests, but they don't generate TOEFL-format questions or drill Listen and Repeat's verbatim-recall demand. Closer to test day, students still need timed practice with real, current-format TOEFL Speaking items.
Written by Prof. Vito Schiuma, designer of English, no kidding.