How Questioning Works in Teaching | Making Thinking Observable

One-Sentence Definition

Questioning in teaching is the deliberate use of prompts, probes and response routines to elicit learner knowledge, reasoning, misconceptions, uncertainty and strategy so teaching can adapt to evidence rather than assumptions.

This page owns questioning as the general mechanism for making thinking observable. How Cold Calling Works owns one respondent-selection routine. How Formative Assessment Works owns the broader use of evidence to adapt teaching. How Teacher Modelling Works owns the exposure of expert thinking. Questioning travels in the opposite direction: it asks the learner to expose their current model so the teacher can decide what should happen next.

The Classroom Where Everyone Says Yes

A teacher finishes explaining the chain rule.

Teacher:

Does everyone understand?

Several students nod. Nobody asks a question.

The teacher moves on.

Homework later contains a common error:

y=(3x+1)⁵
dy/dx=5(3x+1)⁴

The inner derivative is missing across many scripts.

The question “Do you understand?” did not reveal the relevant thinking. It asked students to judge their own understanding globally while the explanation was still fresh.

A better question would have been:

What exactly tells you that this expression needs more than the ordinary power rule?

Now the teacher can hear whether students have built the nested-function schema or are merely following the worked example.

Questions Are Instruments, Not Decoration

A question should have a job.

• activate prior knowledge;
• direct attention;
• retrieve a fact;
• check understanding;
• reveal a misconception;
• test method selection;
• probe reasoning;
• compare alternatives;
• promote self-explanation;
• evaluate confidence;
• test transfer;
• decide whether to progress or reteach.

If the teacher cannot say what the question is meant to reveal, the answer may be difficult to interpret.

The Questioning Control Loop

Choose the thinking you need to inspect → Design the question → Give sufficient thinking time → Gather responses broadly enough → Listen for the model behind the answer → Probe the critical distinction → Correct or extend → Re-sample → Adapt the next teaching move.

Start With What You Need to Know

Suppose the learning target is chain rule.

You might need to know whether students can:

• identify a composite function;
• distinguish it from product structure;
• retrieve the rule;
• differentiate the outer function;
• differentiate the inner function;
• combine both correctly;
• explain why the inner derivative appears.

One broad question cannot diagnose all seven. Questioning works best when the teacher knows which layer is being sampled.

Questions Can Activate Prior Knowledge

Before new teaching:

Differentiate x⁵. Differentiate 3x+1. What happens when these two structures are nested?

The first two questions activate familiar knowledge. The third creates the bridge to the new idea.

See How Prior Knowledge Works.

Questions Can Direct Attention

Students do not always notice the feature experts notice.

Teacher:

Before calculating, what is inside what?

The question directs attention toward nesting rather than the visible exponent alone.

A question can therefore be an attentional scaffold.

Questions Can Trigger Retrieval

Questions are natural retrieval cues.

Examples:

• What is the quadratic formula?
• What does diffusion mean at particle level?
• What does “evaluate” require in this paper?
• What was the central argument of yesterday’s passage?

When learners answer from memory before notes are opened, questioning becomes retrieval practice.

Questions Can Reveal Misconceptions

Some questions should be designed around common wrong models.

Instead of:

Do plants make food?

ask:

A plant grows in soil for months. Where does most of the dry mass added to the plant come from, and why?

The answer exposes whether the learner believes mass comes mainly from soil or can connect carbon dioxide to photosynthesis and biomass.

See How Misconceptions Work.

Questions Can Test Schema Boundaries

Ask learners to compare near neighbours.

Why does (3x+1)⁵ need chain rule but x²(x+1) need product rule?

This question tests whether the learner’s schema is based on deep structure rather than surface features such as brackets.

Questions Can Test Method Selection

Students often practise execution after a worksheet has already announced the method.

A questioning routine can restore selection:

Which rule would you choose here? What feature rules out the nearest alternative?

The second part makes the decision more robust because students must explain the discriminating cue.

Questions Can Reveal Confidence Calibration

Before hearing the answer, ask:

How confident are you, and what evidence supports that confidence?

A student who is highly confident and wrong may have a misconception. A student who is correct but very uncertain may need more retrieval evidence and fluency.

The answer itself and the confidence attached to it provide different information.

Questions Can Probe Reasoning

The first answer often reveals only the conclusion.

Follow with:

• Why?
• What evidence supports that?
• What would make you change your answer?
• Which step is doing the main work?
• What assumption are you making?
• What alternative did you reject?

Probing turns answers into models of thought.

“Why?” Is Powerful but Often Too Broad

Students sometimes freeze when asked only “Why?” because several kinds of justification are possible.

More diagnostic:

Which part of the expression tells you the inside changes with x?

or:

What evidence from the passage makes that inference reasonable?

Specific questions reduce ambiguity while preserving reasoning.

Question Sequences Matter More Than Isolated Questions

Strong teaching often uses a ladder.

1. What do you notice?
2. What does that feature mean?
3. Which schema does it activate?
4. Which method follows?
5. Why not the alternative?
6. What is the first step?
7. What common error might appear?
8. How would you check?

The sequence walks through the expert control system while giving the teacher multiple diagnostic checkpoints.

The Funnel: Broad to Specific

Start broad when you want to see the learner’s natural representation:

What do you notice?

Then narrow:

What is inside what?

Then narrow again:

What is d/dx(3x+1)?

The level at which the student recovers reveals the first weak link.

The Ladder: Specific to General

Sometimes start with an accessible fact and build upward.

1. What is the derivative of 3x+1?
2. Where does that derivative appear in the chain-rule solution?
3. Why must it appear?
4. What general rule can you state about nested functions?

This helps students construct abstraction from secure pieces.

Questions Should Match the Learning Phase

During acquisition:

• identify cues;
• retrieve prerequisites;
• explain worked examples;
• complete steps.

During practice:

• select methods;
• compare cases;
• diagnose errors;
• predict consequences.

During mastery:

• transfer;
• evaluate alternatives;
• justify efficiency;
• explain limits;
• create examples and non-examples.

During exam preparation:

• identify command words;
• retrieve under time;
• choose answer form;
• manage stuck questions;
• check high-risk errors.

Wait Time Is Part of Question Design

A reasoning question followed by half a second of silence is not really a reasoning question for most learners. It becomes a speed-selection device.

Allow time proportional to complexity.

Teacher habit:

ask → pause → watch thinking → then sample.

Thinking time is not dead time. It is where the learning operation occurs.

Questions Should Usually Be Asked Before Naming the Student

If the teacher says “Maya…” before asking the question, other students may stop processing.

Instead:

Everyone: what is the missing factor in this derivative, and why? [pause] Maya?

Now the entire room is cognitively addressed before one response is heard.

Volunteer Answers Are Biased Samples

Volunteers are useful, but they are not a random sample of understanding.

They can overrepresent:

• confidence;
• speed;
• verbal fluency;
• high attainment;
• social comfort.

Use volunteer responses for some purposes, but sample more broadly when you need evidence about the class.

Cold Calling Is One Sampling Tool

Cold calling can broaden participation by selecting students who did not volunteer.

But good questioning still requires:

• thinking time;
• appropriate difficulty;
• respectful feedback;
• partial-answer acceptance;
• psychological safety.

The selection method does not rescue a badly designed question.

Whole-Class Response Can Increase Sample Size

Use:

• mini-whiteboards;
• finger signals;
• short written jots;
• polls;
• simultaneous answer cards;
• digital response systems where appropriate.

Then cold call selected students to explain the reasoning behind patterns you observe.

sample everyone quickly → inspect selected thinking deeply.

Questions Can Be Diagnostic Before Teaching

Before a new topic, ask questions that test prerequisites.

Before chain rule:

• differentiate x⁵;
• differentiate 3x+1;
• identify the inner and outer functions in (3x+1)⁵.

If the second item fails, the teacher knows the inner derivative will become expensive later.

Questions Can Be Formative During Teaching

After modelling one example:

For (2x−5)⁻², what stays the same about the structure and what changes?

The answer tells the teacher whether the learner has abstracted the schema.

See How Formative Assessment Works.

Questions Can Evaluate After Teaching

At the end of the lesson:

Create one example that requires chain rule and one that looks similar but does not. Explain the difference.

This tests retrieval, schema boundaries and transfer more deeply than repeating another routine derivative.

Questioning and Feedback Form a Loop

A question reveals a state. Feedback responds to that state.

question → response → interpretation → feedback → re-question.

The re-question is crucial because it checks whether the correction changed the learner’s model.

Do Not Correct Too Quickly

A wrong answer may contain useful partial knowledge.

Student:

I use chain rule because there are brackets.

Teacher could simply say no.

Better:

Brackets are visible in both chain-rule and product-rule questions. What relationship inside the brackets actually matters?

The follow-up helps the learner refine rather than replace the entire response.

Do Not Accept Correct Answers Too Quickly Either

Correct answer:

15(3x+1)⁴.

Follow-up:

Where did the 3 come from, and what error would appear if you forgot it?

The teacher distinguishes lucky or memorised correctness from understood structure.

Ask for Evidence, Not Just Opinions

In English and humanities-style reasoning:

What do you think?

is often weaker than:

What do you think, and what evidence in the text makes that interpretation defensible?

The second question makes the reasoning standard visible.

Ask for Mechanism in Science

Student:

The reaction gets faster because temperature increases.

Teacher:

What changes at particle level between the increased temperature and the increased reaction rate?

The question forces the causal bridge into the answer.

Ask for Representation in Mathematics

Before calculation:

How could you represent this relationship so the unknown becomes visible?

This can reveal whether the learner reaches automatically for algebra, a diagram, a table or a model—and whether that representation actually fits.

Ask for Error Diagnosis

Show:

dy/dx=5(3x+1)⁴.

Ask:

What part is correct? What part is missing? What misconception might produce this exact answer?

Error questions can teach students to see wrong answers as structured evidence.

Ask for Comparison

Comparison builds discrimination.

What makes these two problems look similar, and what single feature means they require different methods?

This is especially useful when misconceptions arise from surface similarity.

Ask for Prediction

Prediction activates the learner’s current model before the outcome is known.

If I double the inner gradient from 3 to 6, which part of the derivative changes and why?

Prediction reveals whether the learner understands causal structure or only remembers a finished calculation.

Ask for Transfer

Once the learner understands a principle:

Where else would this structure appear? What would it look like if the surface changed?

Transfer questions connect classroom knowledge to future unseen tasks.

Ask for Metacognition

Questions can surface the learner’s control process:

• What was your plan?
• Where did you become uncertain?
• How did you decide to change method?
• What would you check next time?
• What cue will you use without me?

These questions gradually become self-questions.

Questioning and Teacher Modelling

Teacher modelling reveals expert thought. Questioning checks whether the learner can reproduce and adapt it.

teacher thinks aloud → learner predicts → teacher models → learner explains → learner applies.

The alternation keeps modelling from becoming passive theatre.

Questioning and Scaffolding

A question can be the smallest possible scaffold.

Instead of showing the whole next step:

Which quantity should stay unchanged while you apply the outer derivative?

The learner still performs the operation.

See How Scaffolding Works.

Questioning and Fading

Teacher questions can become dependence if they never disappear.

Fade:

What is inside what? → Inside? → learner asks internally → silence.

The final goal is not a student who answers the teacher’s questions forever. It is a student who can ask the useful questions internally.

Questioning and Self-Regulated Learning

Many metacognitive prompts begin externally:

• What is your goal?
• What strategy will you use?
• Is it working?
• What evidence do you have?
• What will you change?

Through repeated use, these become the learner’s own control loop.

Questioning and Psychological Safety

Questioning only produces truthful evidence when learners can expose uncertainty without humiliation.

Teachers should normalise:

• partial answers;
• revised answers;
• “I am unsure because…”;
• mistakes used for comparison;
• thinking time;
• asking for clarification.

If students believe every wrong answer threatens status, they may optimise for silence instead of learning.

Questioning and Processing Speed

Fast answers are not always better answers.

Give adequate wait time and distinguish:

• knowledge;
• retrieval latency;
• verbal formulation speed;
• reasoning depth.

A classroom that rewards only instant response can misread slower but accurate thinkers.

Questioning and Cognitive Load

Questions can reduce or increase load.

Useful sequencing:

one relationship at a time → combine once individual pieces are stable.

Asking a novice to explain every possible implication of a new concept immediately may produce overload rather than insight.

Questioning and Retrieval Practice

Questioning is retrieval practice when students produce knowledge from memory.

But teacher questions should not supply so many hints that retrieval disappears.

Use a cue ladder only after the learner has attempted free recall.

Questioning and Error Correction

Questions can lead the learner back to the first weak link.

Error:

5(3x+1)⁴

Diagnostic sequence:

1. What part is already correct?
2. How many function layers are present?
3. Which layer has not contributed a derivative yet?
4. What is its derivative?
5. Now repair the answer.

The student reconstructs instead of merely receiving correction.

Questioning and Peer Discussion

Questions can move through peers before returning to the teacher.

think alone → compare with partner → identify disagreement → explain reasoning → teacher samples.

This increases the number of students who articulate reasoning rather than leaving discussion to one public speaker.

Questioning and Peer Tutoring

Peer tutors need question frames so they do not simply tell answers.

• What do you notice?
• What do you already know?
• Which step is unclear?
• Why does your method fit?
• How can you check?

The next peer-tutoring page owns this interaction more fully.

Questioning and Small-Group Tuition

Small groups allow higher-resolution questioning because the tutor can inspect each learner’s reasoning frequently.

Three students can answer the same question at different layers:

• Student A identifies the cue;
• Student B explains the method;
• Student C diagnoses a common error.

The teacher can then repair each first weak link quickly.

The Question Bank Should Store Jobs, Not Just Questions

A useful teacher question bank can tag items by purpose:

• prerequisite retrieval;
• misconception detection;
• schema boundary;
• method selection;
• explanation;
• transfer;
• error diagnosis;
• metacognition;
• exam command;

This prevents questioning from becoming a random stream of prompts.

The Hinge Question

A hinge question sits at a decision point in teaching.

The class answer determines whether the teacher can move on.

Example:

Which expression requires the chain rule, and what feature distinguishes it from the others?

If many students select based on brackets rather than composition, progression would build on a weak schema.

The hinge question prevents the next floor from being built too early.

The Exit Question

At the end of a lesson, ask for evidence of independent learning.

Create one chain-rule question, solve it, then explain what makes it chain rule rather than product rule.

Creation plus explanation reveals more than “What did we learn today?”

The Next-Lesson Retrieval Question

Questions should travel across time.

Next lesson:

Without notes, what was the structural cue for chain rule, and what was the common error?

Now the teacher checks retention, not just immediate understanding.

The Questioning Audit

1. What thinking do I need to inspect?
2. Does the question reveal that thinking directly?
3. Does it accidentally contain the answer?
4. Is prior knowledge sufficient?
5. How much thinking time is needed?
6. Who will answer, and how broad is my sample?
7. What wrong answers do I expect?
8. What follow-up will distinguish misconception from slip?
9. How will I respond to partial answers?
10. What teaching decision will the responses control?
11. How will I re-sample after feedback?
12. Will this question eventually become a learner self-question?

The Questioning Traffic Light

• Red: questions are rhetorical, volunteer-biased or used mainly to check compliance—redesign around observable thinking and broader sampling.
• Amber: questions reveal answers but not reasoning—add follow-up on cues, evidence, method and error.
• Green: questions reveal learner models, participation is broad, responses change teaching and useful prompts are transferring into student self-questioning—maintain and deepen.

Questioning in Mathematics

Mathematics questions can inspect:

• representation;
• structure recognition;
• method selection;
• procedure;
• justification;
• error detection;
• checking;
• transfer.

The Mathematics Learning Hub owns the subject content. Questioning reveals which layer of mathematical performance is actually weak.

Mathematics Case: Chain Rule

Question 1:

What do you notice in y=(3x+1)⁵?

Question 2:

What is inside what?

Question 3:

Which rule does that structure activate?

Question 4:

Why is 5(3x+1)⁴ incomplete?

Question 5:

How would the answer change if the inner function were 6x+1?

Each question samples a different layer of the same schema.

Questioning in English Reading

High-value questions include:

• Which word changes the relationship between these sentences?
• Who does this pronoun refer to, and what evidence constrains the answer?
• What can we infer, and which exact detail justifies it?
• What interpretation would go beyond the evidence?
• How does the writer’s language create the effect?

The goal is to make reading processes visible enough to teach.

Questioning in English Writing

Before drafting:

• What is the prompt really asking?
• What is your central claim or story problem?
• What job does this paragraph do?
• What evidence or detail earns its place?
• What should the reader understand by the end?

After drafting:

• Which paragraph is least relevant?
• Where does your reasoning jump?
• Which sentence could be more precise?
• What recurring error should you edit first?

Questioning in Science

Science questioning should move beyond vocabulary recall.

• What changed?
• What mechanism responds?
• Why does that mechanism produce the effect?
• What evidence supports the conclusion?
• What alternative explanation can we rule out?
• What would happen if the condition reversed?
• Where does the model stop being useful?

Questions make causal and evidential reasoning inspectable.

Primary School Questioning

Young learners benefit from short, concrete questions that build toward explanation.

What do you see? What do you know? What changed? Why do you think that happened?

Use pictures, objects and simple examples. Give enough wait time. Accept partial language while preserving conceptual expectations.

PSLE Questioning

P5 and P6 questioning can connect subject knowledge to examination performance:

• What does the command word require?
• What is the first weak link in your solution?
• What quantity is the reference base?
• Which Science mechanism carries the mark?
• What evidence supports your comprehension answer?
• What should you check before moving on?

Secondary School Questioning

Secondary students can be asked increasingly to compare, justify and evaluate.

• Why this method rather than the nearest alternative?
• Which assumption is hidden?
• What evidence would falsify your explanation?
• What changes if one condition is removed?
• Can you construct a counterexample?
• What is the most efficient representation and why?

As expertise grows, questioning moves from recall toward disciplined judgement without abandoning retrieval.

O-Level Questioning

Near O-Levels, questions should become performance-specific:

• What is the examiner asking?
• Which knowledge must be retrieved?
• Which method gives the shortest reliable path?
• What common error is likely here?
• How many marks remain available if the full solution is unclear?
• When should you move on?
• What does the mark scheme reward?

These questions should progressively become the learner’s own internal exam routine.

The Sports Performance Crosswalk

Coaches ask athletes:

What did you see? Why did you choose that option? What changed after the defender moved? What would you do next time?

The questions turn tacit perception into explicit learning and help the coach distinguish technical failure from decision failure.

The Logistics Crosswalk

Operational teams use diagnostic questions to find failures:

• Where did the process first diverge?
• Which input was missing?
• Was the procedure known?
• Was the signal ignored?
• What would prevent recurrence?

Good teaching questions are similarly diagnostic: they locate the first divergence rather than merely record the final failure.

The Governance Crosswalk

Good governance depends on asking questions that reveal inconvenient information rather than questions designed to produce reassurance.

“Is everything fine?” is weak governance.

Which indicator would tell us first that the current plan is failing?

Classroom questioning needs the same preference for informative answers over comfortable ones.

Questioning and AI

AI can generate questions at scale, but quantity is not the same as diagnostic quality.

Strong AI use:

define misconception or skill → ask AI for discriminating questions → teacher verifies them → learner answers before hints → AI or teacher probes → fresh question verifies repair.

Weak use is an endless stream of generic questions with no model of what each answer is supposed to reveal.

Common Failure Mode 1: “Do You Understand?”

Self-report substitutes for evidence.

Repair: ask the learner to produce, classify, explain or apply.

Failure Mode 2: Only Volunteers Answer

The sample is biased toward confidence.

Repair: broaden response through cold calling, written jots or whole-class response.

Failure Mode 3: No Wait Time

Questions measure speed more than thinking.

Repair: give thinking time proportional to complexity.

Failure Mode 4: Teacher Answers Own Question

Students learn that silence will eventually produce the answer.

Repair: wait, scaffold lightly and re-sample.

Failure Mode 5: Correct Answer Ends the Question

Reasoning remains unknown.

Repair: ask why, cue, evidence or alternative.

Failure Mode 6: Wrong Answer Is Rejected Immediately

Partial knowledge and misconception structure disappear.

Repair: identify what is right, then probe the exact divergence.

Failure Mode 7: Questions Contain the Answer

Recognition replaces retrieval.

Repair: remove unnecessary hints and ask open enough to reveal the model.

Failure Mode 8: Questions Are Too Broad

Students cannot tell what kind of reasoning is required.

Repair: narrow to the diagnostic relationship.

Failure Mode 9: Questions Are Too Narrow Forever

Students answer fragments but never integrate.

Repair: combine into explanation and transfer after component understanding stabilises.

Failure Mode 10: Questioning Becomes Public Threat

Students optimise for silence.

Repair: rebuild psychological safety, thinking time and respectful error handling.

Failure Mode 11: Teacher Does Not Use the Evidence

Questions become performance theatre.

Repair: decide in advance what response pattern would trigger reteaching, practice or progression.

Failure Mode 12: Helpful Questions Never Fade

Teacher remains the learner’s metacognitive system.

Repair: transfer useful prompts into student self-questioning.

What Parents Can Ask

• Can my child explain why an answer is correct?
• Can they identify the cue that triggered the method?
• What question helps them notice their common error?
• Can they answer without the notes open?
• Can they compare two similar problems?
• What question should they eventually learn to ask themselves?

What Teachers Can Do

Design questions from the information you need, not from habit. Ask before naming the respondent. Give enough wait time. Sample beyond volunteers. Probe both correct and incorrect answers. Ask for evidence, mechanism and method cues. Use hinge questions before building the next layer. Re-sample after feedback. Fade prompts into student self-questioning. Treat every question as an instrument for making hidden learning visible.

What Tutors Can See in a Small Group

With three learners, questioning can become extremely precise. One student can identify the structure, one can explain why, and one can diagnose a wrong answer. The tutor can then ask a different follow-up to each learner according to the first weak link.

The value of the small group is not simply that everyone gets more turns. It is that the tutor can see more of each learner’s thinking and correct sooner.

Case Study 1: The Nodding Class

A teacher ends explanations with “Everyone okay?” Homework repeatedly reveals the same misconception.

The teacher replaces the question with one hinge item and one explanation prompt. The misconception is visible before independent practice begins, and the lesson branches immediately.

Case Study 2: Chain Rule

Students can calculate chain-rule derivatives but the teacher is unsure whether they recognise the structure independently.

The teacher displays four unlabeled expressions and asks:

Which requires chain rule, and what feature excludes the other three?

Several students choose based on brackets. The teacher now knows the schema boundary needs repair before harder calculus is introduced.

Case Study 3: English Inference

A student gives correct inference answers but cannot explain evidence. The teacher begins asking “What makes that inference defensible?” after every answer.

Some answers collapse; others strengthen. The class learns that inference is constrained reasoning rather than intuition alone.

Case Study 4: Science Mechanism

A class states effects accurately but omits mechanisms. The teacher’s question changes from “What happens?” to “What happens in between?”

That one diagnostic question exposes the missing causal middle and changes later written responses.

Case Study 5: The Volunteer Bias

Five confident students answer almost every question. The teacher introduces written jots followed by deliberate sampling across the class.

Unexpectedly, several quiet students demonstrate strong reasoning while some frequent volunteers reveal shallow but fast answers. The teacher’s model of the class becomes more accurate.

Case Study 6: The Over-Questioned Student

A tutor asks so many guiding questions that a student can solve every problem only in dialogue.

The tutor begins fading: full question → cue word → silent wait. The learner’s first attempts become slower, then more independent. The questions had been useful scaffolds, but they needed to become self-questions.

Case Study 7: The O-Level Hinge

Before starting full differentiation papers, a teacher asks a mixed classification set: ordinary power, product, quotient and chain.

The class can execute each method in isolation but misclassifies two mixed forms. The teacher postpones full timed papers for one lesson and repairs method selection first.

A single well-designed question sequence prevents time pressure from being added on top of an invisible selection weakness.

The Questioning Performance Control Loop

Decide what invisible thinking must become visible → ask a question that samples that exact layer without giving away the answer → allow enough time for genuine retrieval and reasoning → gather responses broadly enough that confidence does not become the sampling system → listen for the schema, misconception or strategy behind the words → probe the first meaningful distinction → respond with the smallest useful correction or extension → ask again on a fresh case → let the response determine the next teaching move → transfer the most useful teacher questions into the learner’s own internal control system.

Canonical Owner Boundaries

This page owns questioning in teaching as the deliberate design, sequencing and sampling of questions used to reveal learner knowledge, reasoning, misconceptions, uncertainty, method selection and metacognitive control so teaching can respond to evidence. It connects to:

• How Cold Calling Works — one specific method for selecting respondents beyond volunteers.
• How Formative Assessment Works — using elicited evidence to adapt instruction.
• How Teacher Modelling Works — exposing expert thinking before questioning learners about the same structure.
• How Scaffolding Works — questions used as temporary prompts at the first weak link.
• How Metacognition Works — teacher questions gradually becoming learner self-questions.

Evidence and Limits

Questioning is a broad family of practices rather than one intervention. Its value depends on question quality, timing, response sampling, learner prior knowledge, classroom culture and what the teacher does with the answer.

Open questions are not always superior to closed questions. A precise closed question can diagnose a prerequisite efficiently, while an open question can reveal richer reasoning. Similarly, higher-order questions are not always appropriate when learners still lack basic knowledge. The question should match the instructional job.

The strongest practical rule is ask for information you can use: decide which part of the learner’s mental model matters next, design the smallest question that exposes it, and let the answer change what happens in the lesson.

The Return Path

Return to the classroom where everyone said yes.

The nods were not lies.

They were simply weak evidence.

Questioning works in teaching when it converts the invisible into the inspectable—when a teacher can hear not only whether an answer is right, but what the learner noticed, retrieved, misunderstood, selected and believed; and when that evidence arrives early enough that teaching can change before the final test makes the hidden problem visible too late.

That is how questioning works in teaching.