English comprehension is a reasoning skill: how to teach inference

A parent tells us their child reads constantly. Novels at bedtime, articles on the bus, a vocabulary well beyond their age. Then the school report arrives and the English comprehension mark is the weakest on the page.

It is one of the most common puzzles we hear, and the explanation is simple. Reading a passage and answering questions about it are different skills. The first is fluency. The second is reasoning, and reasoning can be taught.

Two kinds of question hide in every paper

Almost every comprehension paper mixes two types of question.

Retrieval questions have their answer on the page. “Where did the family live?” “Give two reasons the narrator was late.” A careful reader finds the line and copies or rephrases it.

Inference questions have no answer on the page. The writer has left clues, and the student has to assemble a conclusion from them. “How does the narrator feel about her brother?” “What does the word ‘crept’ suggest about the mood?” “Why do you think the writer ends with a short sentence?”

Strong readers often struggle on inference precisely because, reading for pleasure, they infer without noticing. An exam asks them to make that invisible step visible, and nobody has taught them how.

How an inference question is built

Once you see how examiners write these questions, they stop looking mysterious. Across the six boards we teach (IB, Cambridge International, the US High School Diploma with AP, CBSE, CISCE and NIOS), the construction is remarkably consistent. It takes three moves.

First, the examiner picks a moment where the writer shows rather than tells. Take this sentence:

Maya pushed the letter to the back of the drawer, under the old phone chargers, and went down to dinner smiling a little too widely.

The writer never says Maya is worried. That gap between what is shown and what is meant is exactly where the question will sit.

Second, the question asks about the thing that is not stated. A feeling, a motive, an attitude, an effect. “How does Maya feel about the letter? Use evidence from the text. (2 marks)”

Third, the mark scheme rewards a chain, not a guess. It wants a claim (“she is anxious about the letter and wants to avoid it”), the evidence that supports it (“pushed it to the back of the drawer”, “smiling a little too widely”) and the link that explains how one leads to the other. Hiding the letter under clutter suggests she wants it out of sight. “A little too widely” suggests the smile is forced, so she is pretending to be fine.

That chain is the whole skill. A student who writes “she is sad” may get nothing, even though “sad” is close, because there is no evidence. A student who writes “she puts the letter in the drawer” gets nothing either, because that is retelling, not inferring.

The families of inference question

Most inference questions fall into a handful of families, and naming them helps students recognise what is being asked:

• Character: what someone feels, wants or thinks, but does not say.
• Word choice: what a particular word or phrase suggests beyond its dictionary meaning. “Crept” and “walked” describe the same movement and create very different moods.
• Writer’s purpose and attitude: why the writer included a detail, or what they think about their subject.
• Cause and consequence: why something happened, or what is likely to happen next, based on the clues so far.

Why capable children lose the marks

When we look at a new student’s comprehension answers, the lost marks almost always come from the same four habits:

1. Retelling instead of inferring. The answer describes what happened, not what it means.
2. Answering from life, not from the text. “She feels nervous because letters are usually bad news” is a reasonable thought about the world, but it is not evidence from the passage.
3. Claims that are too strong. “Maya is terrified and is hiding a terrible secret” goes further than the clues allow. Good inference is careful. It claims only what the evidence can carry.
4. Quoting without explaining. The right evidence, dropped into the answer with no link, earns half the marks at best.

None of these is a reading problem. Each is a reasoning problem, and each can be corrected.

How we teach inference

We make the gap visible. Before a student answers anything, we ask them one question: “What is the writer making you work out here?” Naming the gap tells them the answer will not be on the page and has to be built.

We give them a frame. For older students, every inference answer follows the same shape: claim, evidence, link. In practice that sounds like “The writer suggests that… because… which shows…”. It feels mechanical for a few weeks. Then it becomes a habit, and the frame disappears into fluent writing.

We start with pictures and talk for younger children. A seven-year-old can infer brilliantly from a photograph: “Why do you think the boy is holding his arm like that?” “How do you know?” We practise the reasoning out loud, long before we ask for it in writing.

We turn students into examiners. Once a student understands how a question is built, we give them a passage and ask them to write the inference question themselves, along with the mark scheme. Nothing teaches the structure faster than having to build it.

We train multiple-choice elimination. For “which is most likely” questions, we teach students to cross out the options that are true in real life but not supported here, and the ones that go further than the text allows. What is left is usually the answer.

What parents can do at home

You do not need a worksheet. The best practice is conversation.

Ask “how do you know?” about everyday things. About a character in a film (“Do you think he trusts her? What made you think so?”). About an advertisement (“Who is this trying to sell to?”). About a headline (“Is the writer for this or against it?”).

The question that matters is always the second one. “How do you know?” asks a child to point to evidence, and that is the habit every comprehension paper is quietly testing.

Reading between the lines can be learned

Inference is often treated as a gift that some children have and some do not. In our experience, the children who “just get it” have simply made the reasoning automatic. The ones who struggle have not yet been shown that there is a reasoning step at all.

Show them the step, give them a frame for it and practise it in conversation, and comprehension stops being the weakest mark on the report.

At TutorWyz, we teach Mathematics, Physics, Chemistry, Biology and English to students on every continent. If your child reads well but loses marks on comprehension, the first assessment is free, and it is usually the fastest way to see exactly where the reasoning breaks down.