Quick read: Children can have preferences, strengths and different prior knowledge, but the evidence does not justify teaching each child mainly through a fixed “visual”, “auditory” or “kinesthetic” learning style. A stronger approach is diagnose the learning problem → choose the representation that best exposes the content → practise → retrieve → vary → return to normal-form performance.
This page replaces an older learning-styles explanation. The original intention—to adapt teaching to the learner—is worth keeping. The mechanism needed correction.
A major review of learning-styles research found no adequate evidence base for matching instruction to a supposed fixed learning style, and the Education Endowment Foundation similarly cautions against labelling pupils by such categories. A 2024 meta-analysis found effects too small, inconsistent and methodologically weak to justify widespread matching. Preferences can exist; that is different from showing that students learn best when teaching is matched to a fixed style.
Useful references: Pashler et al., Learning Styles: Concepts and Evidence and EEF: Learning Styles.
The useful idea hidden inside the myth
The weak idea is “this child is a visual learner, so teach them visually”. The useful idea is “this child is not yet understanding this concept; what representation will make the structure easier to see?”
That shifts attention from identity labels to the actual learning problem.
Preferences are real; fixed instructional prescriptions are the problem
A child may prefer diagrams, discussion, quiet reading or hands-on activity. Preferences can affect engagement and comfort. But a preference does not automatically tell us the most effective form for learning every kind of content.
Some content is inherently visual, some linguistic, some spatial, some procedural. The representation should serve the material and the learner’s current difficulty.
Representation 1: text
Written text is useful when students need to inspect exact wording, reread, annotate, compare sentences or work independently at their own pace.
Text is especially important in English because the final task often requires reading and writing in textual form. Even if another representation helps first, students eventually need to return to text.
Representation 2: spoken explanation
Speech can help when a learner benefits from hearing phrasing, intonation, stress or a teacher thinking aloud through a reasoning process.
But a spoken explanation should not become a permanent substitute for reading. After understanding, the learner should still be able to perform the English task in its ordinary form.
Representation 3: diagram or visual structure
A diagram can expose relationships that are difficult to hold in working memory.
- timeline for tense;
- cause → effect chain for comprehension;
- paragraph map for writing;
- claim → evidence structure for argument;
- character relationship map for narrative reading.
The visual works because it represents the relationship well, not because the student has been classified as a “visual learner”.
Representation 4: worked example
Worked examples reduce unnecessary search when a learner is new to a process.
For example, a teacher can show how to answer an inference question:
- identify the question;
- locate the evidence;
- explain what the clue suggests;
- construct the answer;
- check whether the conclusion goes beyond the text.
Then support should fade.
Representation 5: comparison
Side-by-side comparison is powerful for language because students can see what changes meaning.
- strong versus weak paragraph;
- literal versus figurative wording;
- supported versus unsupported inference;
- precise versus vague vocabulary;
- coherent versus incoherent sentence sequence.
The comparison creates discrimination—the learner sees what feature matters.
Representation 6: enactment or manipulation
Sometimes physically rearranging sentence strips, acting out a dialogue or sequencing event cards makes a relationship easier to understand.
Again, this is not proof of a “kinesthetic learner”. It is a temporary representation that may make an abstract structure concrete.
Worked lab: one problem, several representations
Suppose a Primary student keeps mixing past events in a narrative.
- Text: highlight the verbs in a short paragraph.
- Timeline: place events on a visual line.
- Speech: ask the child to retell the sequence aloud.
- Manipulation: reorder event cards.
- Normal form: rewrite a fresh paragraph with no timeline visible.
- Delayed return: revisit tense several days later in an unseen composition.
The representation is successful only if it improves later independent performance.
The final task should control the destination
If the learner must ultimately read an unseen passage independently, teaching should eventually return to independent reading. If they must speak spontaneously, they must eventually practise spontaneous speech. If they must write, they must eventually write without the diagram or sentence frame.
Support can change the route. It should not quietly change the destination.
Misconception: “My child is visual, so audio teaching will not work”
This can unnecessarily restrict learning.
Repair: ask which representation best fits the content and current bottleneck. A student may benefit from a diagram for one concept and oral modelling for another.
Misconception: multimodal teaching is automatically better
More modes can help when each contributes useful information. They can also overload attention if the learner is asked to process unnecessary text, animation, audio and decoration at once.
Repair: add representations because they clarify the learning structure, not because variety itself is the goal.
Misconception: engagement proves learning
A child may enjoy a game or video without retaining the target skill.
Repair: after the engaging activity, require retrieval or transfer in a new task.
Misconception: preferences should be ignored completely
That also goes too far. Motivation, accessibility, prior experience and comfort matter.
Repair: respect preferences where practical, but do not let them become fixed labels that prevent students from developing less-preferred but necessary capabilities.
A better adaptive-teaching loop
- Observe: what is the learner failing to understand or perform?
- Diagnose: knowledge, language, representation, memory, reasoning or execution?
- Choose: select the simplest representation that exposes the structure.
- Model: make the reasoning visible.
- Practise: student performs with support.
- Fade: remove the representation or prompt.
- Retrieve: test after a delay.
- Transfer: use unfamiliar normal-form work.
Parent/tutor checklist
- What exactly is the child finding difficult?
- Why was this representation chosen?
- What feature is it making easier to see or hear?
- When will the scaffold be removed?
- Can the child perform without it?
- Can they retrieve the skill later?
- Can they transfer it to unseen work?
This is stronger personalisation than assigning an identity label.
How to measure whether adaptation worked
- Did understanding improve?
- Did the learner require fewer prompts?
- Can they explain the concept in their own words?
- Can they perform in the ordinary task format?
- Does the skill survive after a delay?
- Does it transfer when surface features change?
What success looks like
The goal is not to discover the child’s one true learning style. It is to build a learner who can use different representations intelligently and eventually perform without them.
The durable loop is diagnose → represent → understand → practise → fade → retrieve → transfer.
Related routes
For how eduKateSG defines meaningful individual adaptation, see What Personalised Attention Should Actually Mean in Small-Group English Tuition. For progress evidence, see How to Measure Primary English Progress Without Relying on One Score.