Academic English Hub

Editorial policy

Where our facts come from, how our practice material is written and checked, what AI does and does not do here, and how we fix mistakes.

Last updated:

Where our facts come from

Anything on this site that states how the TOEFL works, whether that is section timings, question counts, task structures, or how scores are reported, comes from ETS's own published documentation. Those are the only authoritative sources for exam format, and we treat them as such.

Every guide page carries a sources block listing the specific ETS pages behind its factual claims, together with the date we last checked them. If a page tells you the Reading section has two modules of particular lengths, you can follow the link and confirm it rather than taking our word for it.

We separate two kinds of statement deliberately and try to keep the distinction visible in the writing. Facts about the exam are attributable to ETS. Advice about how to prepare is ours, drawn from the structure of the test and from what we see in practice results, and you should weigh it as opinion.

How practice material is written

Every practice item on this site is written by us. We do not reproduce official ETS questions, published practice sets, or material from any preparation book, and we do not adapt them either, because a lightly reworded passage is still someone else's work.

What we take from official documentation is the specification: how long a passage runs, how many gaps a Complete the Words paragraph contains, which question types appear in a set, how many sentences a Listen and Repeat scenario has and how long you get for each. We then write to that specification on topics of our own choosing.

This matters for two reasons beyond copyright. Practice material you have encountered before measures recall rather than ability, so reused questions produce inflated and useless results. And material written to the current specification stays useful when the exam's surface details change, because the specification is what we track.

  • Passages, prompts, questions, and explanations are original.
  • Audio is produced by us for our own scripts.
  • Topics are chosen by us, not borrowed from any published set.
  • Format specifications come from ETS and are cited.

How material is checked

New items are written against the schema for their task type, which encodes the structural requirements: lengths, counts, answer formats, timing. Validation runs before anything ships, so an item that does not match its task's structure cannot reach a learner.

Beyond structure, items are checked for a single defensible answer, distractors that are wrong for a reason rather than merely odd, and language appropriate to the level. Questions that depend on outside knowledge, or that can be answered without engaging with the material, are rejected. Both are common failure modes in generated practice material.

Difficulty is monitored after release as well as before. Items that almost everyone answers correctly, or almost nobody does, get looked at again: the usual cause is an ambiguity in the wording rather than genuine difficulty.

What AI does here, and what it does not

We use AI in two places and we would rather say so plainly than have you infer it.

The first is authoring support. AI assists in drafting practice material, which is then validated against the task schema and reviewed before release. It is a drafting tool operating inside a specification, not an unattended pipeline. Nothing reaches a learner because a model produced it.

The second is evaluation. Writing and speaking responses can optionally be sent for AI feedback, which returns a practice band with comments on task coverage, language use, and development. This is genuinely useful for spotting patterns across several attempts, and it is genuinely not a TOEFL score. It is not produced by ETS raters, it is not calibrated against official rating, and it should not be used to predict a real result. Where it disagrees with a teacher, trust the teacher.

  • AI assists authoring; validation and review gate what ships.
  • AI evaluation is optional and clearly labelled as an estimate.
  • AI feedback is never presented as an official or predicted TOEFL score.
  • No learner writing or audio is used to train models.

Reporting and correcting errors

We expect to have errors. Practice material at this volume always does, and the useful question is how quickly they get found and fixed rather than whether they exist.

Every question carries a report control on its Help screen. Reporting sends the identifiers needed to locate the exact item, meaning the test, task, and question you were looking at, along with a reason and anything you want to add. It does not pause your test, change your answer, or move you anywhere.

Reports are reviewed against the item itself. A confirmed error is corrected or the item is withdrawn. If a withdrawn item affected a score you have already been shown, that result stays in your history as it was: silently rewriting past results would make your own progress record untrustworthy, which is worse than an occasional stale number.

Keeping up with a changing exam

The TOEFL changed significantly in January 2026, and a transition period around score reporting runs into 2028. Exam formats move, and material that was accurate when written can quietly stop being so.

Guide pages carry the date their factual claims were last verified against ETS sources. When ETS publishes a change, the affected guides and any practice material built on the old structure are updated together. We would rather a page be briefly incomplete than confidently wrong.

If you spot something on this site that no longer matches ETS's current documentation, please tell us. That is the single most useful correction anyone can send.

Advertising and editorial independence

The site carries advertising in one place: on completed practice results, below your score and its explanation, and above the detailed review. Never during a test, never on a question screen or timer, never on sign-in or account pages.

Advertisers have no influence over anything editorial. They do not see your responses, your scores, or what you practised, and no advertiser has any input into what our guides say or how our practice material is written or scored.

Frequently asked questions

Are your TOEFL practice questions copied from official material?

No. Every item is written by us. We take format specifications from ETS documentation, such as lengths, counts, and question types, and write original material to them on our own topics.

Do you use AI to write your practice questions?

AI assists with drafting, but every item is validated against its task schema and reviewed before release. Nothing reaches a learner simply because a model produced it.

Is the AI feedback on my writing or speaking an official score?

No. It is an estimate from our own system, not produced or calibrated by ETS raters. It is useful for finding patterns across attempts and should not be used to predict an official result.

How do I report a mistake in a question?

Use the report control on that question's Help screen. It sends us the identifiers needed to open the exact item, without interrupting your test.

How do you keep the material current?

Facts are checked against ETS documentation and each guide carries the date it was last verified. When ETS publishes a change, affected guides and practice material are updated together.

Does advertising influence your content?

No. Advertisers have no input into our guides, practice material, or scoring, and receive no information about what you practised or how you performed.

Sources and editorial method

Facts that can change were checked against the following official ETS pages on July 24, 2026. Explanations and study advice are original Academic English Hub editorial content.