Infer, Imply, Suggest: What Exam Questions Ask

Exam questions are written in vocabulary too. Words like infer, imply, suggest, primarily, and EXCEPT are instructions, and each rules out a different kind of correct-sounding answer. Most students study the words in the passage and never study the small, recurring set of words that ask the questions.

The half of the test nobody studies

There is a familiar experience in a reading section. You understand the passage. You read the four choices and two of them are obviously wrong. The remaining two both look true, you pick one, and you get it wrong. Afterwards you read the explanation and it says something like “the question asks what the author suggests, not what the author states.”

That is not a comprehension failure. You understood the text. You misread the instruction, and the instruction was one word long.

Every standardized reading test runs on a small stock of question words that recur across forms and across exams. They are not decorative. Each one defines the relationship the correct answer has to the passage, and the wrong answers are built to be attractive to someone who ignored that relationship. A student who reads every stem as a vague “what does the passage say” is answering a different question from the one on the screen.

The strange part is how invisible this vocabulary is. Nobody sells a word list of exam question stems. Prep books teach question types: inference questions, detail questions, vocabulary-in-context questions. That is the same information, organized so that the words themselves never get looked at directly.

Infer, imply, and suggest ask for the same move

These three do most of the work, and students often treat them as three different tasks. They are one task described from different angles.

A writer implies. A reader infers. The word choice depends on which end of the sentence the exam is standing at, not on what you are supposed to do. “What does the author imply” and “what can be inferred from the passage” are the same question. Suggest is the softer synonym and behaves identically on a test.

What all three have in common is more useful than the distinction: they promise that the answer is not written in the passage. It has to be supported by the passage, and the support has to be tight enough that a reasonable person could not read it the other way. But if you can point to a sentence that says it, you have almost certainly found a trap.

This is why the single most common wrong answer on inference questions is a sentence lifted nearly verbatim from the text. It feels safe. It is checkable. It is also, given the stem, disqualified. A correct inference had to travel at least one step from what was printed.

The other trap sits at the far end. Having learned that the answer is not stated, students start choosing answers that require three or four steps of reasoning, some outside knowledge, and a generous assumption about what the author probably believes. Inference on an exam is a short step, not a leap. If you needed a fact the passage never mentioned to get there, you have gone too far.

”According to the passage” is the opposite instruction

Then there is the stem that means exactly what inference stems deny. “According to the passage,” “the passage states,” “the author mentions.” These want something the text says, and the correct answer will be a paraphrase of a specific line you can put your finger on.

The two families are near-mirror images, which is why they get confused under time pressure:

Stem word	What it wants	The wrong answer that tempts you
infer, imply, suggest	Supported, not stated	A near-quotation from the passage
according to, states, mentions	Stated, paraphrased	Something true but not in the text
primarily, mainly, chiefly	The dominant purpose	A real but secondary detail
best, most nearly	The closest fit, not a perfect one	A choice you rejected for being imperfect
EXCEPT, NOT, LEAST	The one that fails	Any of the three that succeed

Read that middle column as a set of promises the test is making. It tells you what a correct answer has to look like before you have read a single choice, which is the whole practical value of noticing the stem.

The qualifiers that quietly decide the answer

Some stem words do not name a task at all. They set a threshold, and they are easy to read straight past because they are ordinary English.

Primarily, mainly, and chiefly concede in advance that other answers will be true. A question asking what a paragraph is primarily concerned with is telling you that two or three choices describe something the paragraph genuinely does. You are ranking, not filtering. Students who look for a true statement find one immediately and stop.

Best and most nearly lower the bar on purpose. “Which choice best describes” and “most nearly means” are both admitting that no choice is exact. The answer does not have to be the word you would have written; it has to beat the other three. This matters enormously on vocabulary-in-context items, where the correct choice is often a slightly awkward synonym that nevertheless fits the sentence, and the tempting wrong choice is the word’s most common dictionary meaning.

Most likely and probably mark the answer as an interpretation rather than a fact, which is another way of signalling inference.

The pattern across all of them is the same: the qualifier tells you that more than one choice will be defensible and that your job is comparison. If you are still asking “is this true?” instead of “is this the most true?”, the qualifier has already beaten you.

EXCEPT, NOT, and LEAST invert the whole question

Negative stems are the only ones the test bothers to shout at you in capitals, which tells you how often people miss them anyway.

On the TOEFL these are formally called negative factual information questions, and they survived the 2026 redesign, appearing in both the academic passage task and the everyday reading task. They look like “the author mentions all of the following EXCEPT” or “according to the paragraph, which of the following is NOT true.”

The mechanics are worth stating plainly, because the format inverts your usual habit. Three of the four choices are correct statements about the passage. You are hunting for the odd one out, which may be a claim the passage contradicts or simply one it never makes. Both count as the answer.

The failure mode is not misunderstanding. It is momentum. You have answered nine questions by looking for the true choice, you find a true choice, and your hand moves before the EXCEPT registers. The capitals exist because the test writers know this and are, in their way, trying to help.

A related trap lives in IELTS reading, where TRUE / FALSE / NOT GIVEN turns a two-way decision into a three-way one. FALSE means the passage contradicts the statement. NOT GIVEN means the passage is silent. Candidates who collapse the two, reasoning that if it is not confirmed it must be false, lose marks on a distinction that is purely a matter of what two words mean.

The verbs that describe what a writer is doing

Beyond the stems themselves, a second layer of vocabulary appears inside the answer choices on harder tests, especially GRE, LSAT-style reasoning, and the upper end of TOEFL and SAT. These describe rhetorical moves rather than content.

• assert, to state something without offering proof
• concede, to grant a point to the other side
• qualify, to narrow a claim without abandoning it
• undermine, to weaken, usually indirectly
• refute, to disprove outright, a far stronger claim than undermine
• characterize, to describe in a particular light
• illustrate, to give an instance of something already claimed
• anticipate, to raise an objection before someone else can

These are precise, and the exam uses them precisely. A passage where the author says “this is true, though only in cities” has qualified a claim. It has not conceded, refuted, or undermined anything. A choice saying the author refutes the opposing view will be wrong if all the author did was weaken it.

Most of these sit in the Academic Word List, the 570 word families that recur across academic texts regardless of subject. That overlap is not a coincidence. Exams are written in academic English, so the language describing the passage comes from the same well as the passage.

Why these words slip past your studying

There are two reasons this vocabulary goes unstudied, and they reinforce each other.

The first is that you already know them. You have read the word suggest ten thousand times. Nothing about it announces itself as a term of art, so it never lands on a list of words to look up. It is only inside a question stem that it carries a technical instruction, and that context is exactly where you are too busy to notice.

The second is that stems live in the questions, and questions are what you do, not what you study. Review a practice test and your attention goes to the passage you misread and the words in it you did not know. The stem is the frame around the mistake and it becomes invisible for the same reason a picture frame does.

This is a close relative of a problem we have written about before: on the digital SAT’s Words in Context questions, all four choices are words you know, and knowing them is not the skill being tested. The stem vocabulary has the same shape. Familiarity is doing the opposite of what it should. It keeps you from studying something you have not actually learned.

How to study a question word so it holds under time

The practical move is to promote these words from background to foreground, and to learn each as an instruction rather than a definition.

Go through the practice tests you have already taken and collect the stems. Not the question types. The actual words. You will find the list is short, that it repeats heavily, and that a handful of words account for most of the questions you got wrong.

For each one, write down three things: what relationship the answer has to the passage, what a tempting wrong answer looks like, and one real example from a test you sat. That third item does more work than the other two, because a rule you have seen fail on your own answer sheet is a rule you remember.

Then review them on a schedule rather than in one sitting. This is ordinary vocabulary work and it decays like ordinary vocabulary. A distinction you understood clearly in September will be soft by November unless something brings it back.

The words in the last two sections are worth genuine study, not just recognition: refute against undermine, concede against qualify. Vocaby carries all of them with IPA, audio, and plain-English definitions among its 29,000-plus entries, and reviews are scheduled with FSRS so a distinction you half-remember comes back before it fades rather than after. If you would rather look one up than drill it, the word explorer is the place to start.

What to actually do

• Read the stem twice before the choices. It costs about a second and it tells you what shape the answer has to be.
• Circle the qualifier. Primarily, best, most nearly, and most likely all mean you are ranking rather than filtering.
• Treat capitals as a stop sign. EXCEPT, NOT, and LEAST invert the question and momentum is what beats you, not comprehension.
• On inference questions, distrust the choice you can point to in the text. Supported and stated are different things.
• Build a stem list from your own wrong answers and review it on a spaced schedule alongside your passage vocabulary.

The two vocabularies on a reading test are not equally sized. The passage can throw thousands of words at you and there is no way to prepare for all of them. The questions draw on a set small enough to learn in a week, and every one of those words is guaranteed to appear on test day.

Learn English vocabulary with Vocaby on the App Store