Quick Read
The main problem with teaching English may be that we treat English as though it were only one school subject.
It is not.
In school, English can also function as the language through which students are told:
- what to learn;
- what a question means;
- what operation to perform;
- which information matters;
- what conditions apply;
- how evidence should be interpreted;
- how an answer should be organised;
- and what counts as an acceptable response.
A Mathematics student may know the mathematics and still misunderstand the question.
A Science student may know the concept and still fail to distinguish describe from explain.
A Humanities student may know the historical content but misunderstand what infer, compare, useful, reliable or extentrequires.
And an English student faces an even stranger problem:
English is often taught through English.
If the language used to explain the language is already beyond the student’s working control, the teaching itself can become difficult to decode.
At eduKateSG, we therefore use a systems metaphor:
English is not only a language.
In education, it can also act like a command-and-directives compiler.
Not literally a computer compiler.
Rather, the learner repeatedly has to convert linguistic instructions into executable intellectual operations:
language received
→ meaning parsed
→ task reconstructed
→ operation selected
→ knowledge activated
→ answer produced
→ answer checked.
If the error occurs near the beginning, everything downstream can be wrong even when the student possesses substantial subject knowledge.
That is the problem this article investigates.
The Main Problem With Teaching English
We usually organise education into subjects.
English.
Mathematics.
Science.
History.
Geography.
Literature.
This administrative separation is useful.
But cognitively, the boundaries are much less clean.
A Mathematics question may contain:
conditions, quantities, relationships, exceptions and instructions.
A Science question may contain:
observations, causal relationships, qualifications, comparison and evidence.
A Humanities question may contain:
sources, claims, perspectives, temporal relationships and competing interpretations.
Before the student can do the Mathematics, Science or Humanities, the student often has to understand what the language is telling them to do.
That produces an important distinction:
Subject problem ≠ Language-of-the-subject problem.
Sometimes they occur together.
Sometimes they do not.
And if we fail to separate them, tuition can repair the wrong thing.
English Is a Meaning System
Singapore’s Ministry of Education describes language as a means of making meaning and communication, and as a system with rules and conventions. Its English syllabus also stresses the importance of purpose, audience, context and culture in language use.
That sounds straightforward.
But its educational implications are enormous.
Consider what happens whenever a teacher writes:
Explain why the temperature increased.
The sentence is not merely carrying vocabulary.
It contains an instruction.
The student must determine:
What object am I discussing?
Temperature.
What happened?
It increased.
What operation am I required to perform?
Explain.
What kind of relationship is required?
Causal.
So before Science knowledge can be deployed, language has already constructed the task.
English has told the learner what kind of cognitive operation must occur.
The Command-and-Directives Compiler
This is why we use the term Command and Directives Compiler as an eduKateSG educational metaphor.
A computer compiler converts instructions written in one representation into something executable by another system.
Human language processing is vastly more complicated and should not be confused with computer compilation.
But the metaphor is useful.
In school, students repeatedly receive something like:
Compare X and Y and explain which factor was more significant.
That sentence has to be transformed into an internal operating plan.
Something like:
identify X
→ identify Y
→ establish dimensions for comparison
→ find similarities/differences
→ establish criteria for significance
→ evaluate evidence
→ make judgement
→ justify judgement.
The final answer may be History.
But the command parser is language.
If the learner interprets “compare” as:
write everything I know about X and then everything I know about Y,
the historical knowledge may be correct while the answer remains poorly aligned to the task.
The student does not necessarily have a History problem.
The student may have a directive-decoding problem.
One Word Can Change the Entire Operation
Consider these instructions:
Describe the change.
Explain the change.
Compare the changes.
Suggest a reason for the change.
Evaluate the importance of the change.
The topic can remain identical.
The intellectual operation does not.
Now consider:
State one reason.
versus:
Explain one reason.
Or:
What happened?
versus:
Why did it happen?
Or:
To what extent do you agree?
versus:
Do you agree?
Tiny changes in language can alter:
- depth;
- scope;
- evidence requirements;
- reasoning;
- organisation;
- degree of judgement;
- and the expected answer.
That makes English an unusually important layer of educational infrastructure.
Before Solving the Problem, the Student Must Build the Problem
This is particularly visible in Mathematics word problems.
A student sees:
Sarah has 24 fewer stickers than Amir.
The mathematics may be elementary.
But before calculating anything, the learner must construct the relationship correctly.
Who has more?
Who has fewer?
What does the number represent?
Is 24 a total?
A difference?
An amount to subtract?
What quantity is unknown?
Researchers studying mathematical word problems increasingly separate text comprehension and mathematical computation because successful word-problem solving requires the learner to construct the correct problem representation before calculating. A 2025 study using eye movements found that text comprehension preceded computation in word-problem solving.
That gives us a useful educational chain:
language
→ situation model
→ mathematical model
→ computation.
If the situation model is wrong, perfect arithmetic can produce the wrong answer.
A Mathematics Error May Begin Before the Mathematics
Consider:
Ali has three times as many marbles as Ben.
A student may know multiplication perfectly.
But what if the relationship is reversed?
Or:
The tank is filled to three-fifths of its capacity.
The fraction concept may be understood.
But what if “capacity” is misunderstood?
Or:
At least 12 students must attend.
“At least” establishes a boundary.
It does not mean exactly 12.
Similarly:
no more than,
respectively,
remaining,
difference between,
per,
consecutive,
increases by,
increases to,
proportional to,
unless,
if,
only if
all carry relationships.
Mathematics therefore contains a language interface wrapped around mathematical structures.
This does not mean mathematics is merely English.
It means that when mathematics is delivered through English, inaccurate language decoding can interfere with access to the mathematical problem.
Science Has Its Own Language Layer
Science presents another form of the same problem.
Scientific language tends to demand precision.
Students meet distinctions such as:
mass versus weight;
heat versus temperature;
accuracy versus precision;
observation versus explanation;
correlation versus causation;
energy versus force.
Everyday language sometimes tolerates fuzziness that scientific language cannot.
Research on science education has long identified academic language as a substantial challenge because scientific writing compresses complicated relationships into precise linguistic structures.
A student may therefore have several simultaneous jobs:
understand the vocabulary
→ parse the sentence
→ identify the scientific relationship
→ retrieve relevant scientific knowledge
→ construct a causal model
→ formulate an answer using acceptable scientific language.
If the language layer is fragile, Science becomes harder even before the underlying scientific concept is considered.
Humanities Adds Perspective and Evidence
Humanities increases the complexity again.
The student may have to distinguish:
fact;
claim;
evidence;
inference;
perspective;
purpose;
reliability;
usefulness;
consequence;
significance;
comparison;
continuity;
change.
The student is not merely retrieving content.
The student is being asked to perform operations on information.
This is why two learners who memorised the same chapter can achieve very different results.
One possesses information.
The other can also understand the directive controlling what should be done with that information.
So English Sits Under More of School Than We Usually Notice
We can represent this approximately as:
English / Language Interface
feeding into:
Mathematics
Science
Humanities
Literature
examinations
worksheets
textbooks
teacher explanations
feedback
research
online learning.
Not every school subject depends on English to the same degree.
And obviously, conceptual, quantitative, spatial, procedural and domain-specific abilities remain independently important.
But language frequently becomes one of the interfaces through which those capabilities are accessed and demonstrated.
That makes weak English potentially multiplicative rather than merely additive.
The Dangerous Part: Errors Can Propagate
Suppose a student misunderstands one directive.
A possible chain is:
question misunderstood
→ wrong task model
→ wrong knowledge retrieved
→ wrong operation performed
→ weak answer
→ low mark.
Now something more dangerous can happen.
The low mark is interpreted as:
“Weak Science.”
So the student receives more Science notes.
But the original problem was:
inability to distinguish describe from explain.
The Science content increases.
The decoding problem remains.
Another examination arrives.
The student misinterprets the instruction again.
Another weak result.
Now everyone becomes increasingly convinced that the student has a Science problem.
We have created a diagnostic illusion.
The Error Can Travel in the Opposite Direction Too
We should be equally careful not to blame every problem on English.
A student can understand the instruction perfectly and simply not know the Science.
Another can understand every word in the Mathematics problem but not understand ratios.
Another can understand the History source but lack the historical context required to interpret it intelligently.
So:
language weakness ≠ universal explanation.
The point is diagnostic separation.
Ask:
Did the learner understand the words?
Then:
Did the learner understand the relationships?
Then:
Did the learner reconstruct the task correctly?
Then:
Did the learner possess the required subject knowledge?
Then:
Could the learner perform the required operation?
Then:
Could the learner communicate the result?
Only then do we know where the failure actually occurred.
Alien 1: The Student Meets the Finished Artefact
Our newest education research produced a useful thought experiment.
Imagine an alien encountering:
a² + b² = c²
for the first time.
We called this Alien 1.
Alien 1 sees the finished object.
The formula is already compressed.
Elegant.
Efficient.
Powerful.
But from the viewpoint of a complete outsider, almost everything that humans have compressed into it is missing.
What is a?
What is b?
What is c?
Why squares?
Under what conditions?
What geometrical object is involved?
Why should anyone care?
What would count as proof?
What problem does this relationship solve?
Alien 1 faces the problem many students face in school.
They encounter the finished representation.
Then education tells them what to do with it.
English Teaching Has Its Own Alien 1 Problem
Consider a student being taught:
Use sophisticated vocabulary.
Or:
Write a good introduction.
Or:
Use PEEL.
Or:
Show, don’t tell.
Or:
Provide textual evidence.
Or:
Make an inference.
Or:
Elaborate.
Or:
Improve your sentence structure.
These instructions can become compressed educational artefacts.
The expert understands the hidden system.
The novice may not.
“Make an inference” sounds like three simple words.
But underneath it may be:
locate textual clue
→ activate relevant knowledge
→ identify an unstated relationship
→ generate plausible interpretations
→ eliminate interpretations contradicted by text
→ choose best-supported interpretation
→ express precisely.
The student receives:
infer.
The teacher carries the entire hidden architecture.
This is Alien 1.
The Finished Essay Has the Same Problem
Give a student an excellent model essay.
It may be:
coherent,
precise,
elegant,
well-supported,
well-organised.
The learner sees the output.
But the output does not automatically reveal:
how the writer generated the ideas;
what alternatives were rejected;
how the argument was organised;
which background knowledge was activated;
how examples were selected;
how claims were qualified;
which sentences were rewritten;
which weak ideas were deleted;
how the conclusion emerged.
The student sees the product.
Not necessarily the production system.
This is one of the central problems with teaching through model answers.
Alien 2: Reconstruct the World That Produced the Answer
Our second thought experiment changes the observer.
Alien 2 does not simply encounter the finished Pythagorean formula.
Alien 2 encounters something closer to the problem-world from which such knowledge could emerge:
measurement,
construction,
shape,
right angles,
length,
area,
practical problems,
regularities,
verification,
proof.
Now the formula is not merely decoded.
Its genesis can be reconstructed.
This leads to one of the strongest conclusions from the newer education research:
Teaching is not merely the delivery of finished knowledge.
Teaching reconstructs the path by which the knowledge becomes usable.
That idea has major implications for English tuition.
What Would Alien 2 English Tuition Look Like?
Instead of giving:
“Use better vocabulary,”
we reconstruct why vocabulary is needed.
The student has an internal distinction.
Current vocabulary cannot encode it precisely.
So:
idea resolution
→ lexical search
→ candidate words
→ meaning check
→ register check
→ grammatical fit
→ final selection.
Instead of:
“Write a better paragraph,”
we reconstruct the paragraph’s job.
claim
→ support
→ relationship
→ evidence/example
→ interpretation
→ landing.
Instead of:
“Make an inference,”
we reconstruct inference.
evidence
- relevant knowledge
- constrained reasoning
→ interpretation.
Instead of:
“Answer the question,”
we reconstruct the directive.
question verb
→ object
→ scope
→ condition
→ required operation
→ evidence standard
→ answer architecture.
Now the student does not merely imitate a result.
The student begins acquiring the capability that generates the result.
This May Be the Main Problem With Education More Broadly
Education has to compress knowledge.
It has no choice.
Human civilisation has accumulated too much knowledge for every student to rediscover everything from first principles.
We therefore use:
textbooks,
definitions,
formulae,
rules,
diagrams,
models,
rubrics,
worked solutions,
categories.
Compression is powerful.
It allows enormous bodies of knowledge to travel quickly.
But compression creates a danger.
The expert sees the compressed object plus the hidden architecture.
The novice may see only the compressed object.
If teaching repeatedly delivers compressed outputs without reconstructing enough of the architecture underneath, students can become good at manipulating symbols they do not deeply control.
English Can Magnify This Problem Because English Is Used to Explain the Compression
This is where teaching English becomes unusually difficult.
Imagine a student does not understand the word:
contrast.
The teacher explains:
“To contrast means to examine the differences between two or more things.”
But suppose the student is also uncertain about:
examine,
differences,
between,
two or more.
The explanation itself contains another decoding problem.
We encounter recursion:
English is used to explain English.
A dictionary uses words to explain words.
A grammar rule uses grammar to explain grammar.
Teacher feedback uses language to explain why the student’s language did not work.
A comprehension strategy is explained using language to a student whose comprehension is weak.
This is not impossible.
Human beings learn language this way successfully every day.
But it means teachers need to remain aware of the instruction-language floor.
If our explanation sits above the learner’s current decoding capability, more explanation can sometimes produce more confusion.
The Student Can Become an Alien Inside the Classroom
This gives Alien 1 another useful interpretation.
Imagine walking into a classroom where everyone appears to understand an invisible code.
The teacher says:
“Evaluate the author’s effectiveness in conveying his attitude.”
Other students begin writing.
You recognise the words individually.
But you cannot reconstruct what operation the sentence requires.
From the outside, it may appear that you:
are inattentive,
did not revise,
lack ideas,
are careless,
or are weak academically.
Internally, something else happened.
The interface never compiled.
This is educationally important.
Because a student can appear to be failing at the task while actually failing at the translation into the task.
Voynich: Is the Interface Actually Intelligible?
Our Voynich work provides another useful educational test.
When confronted with an unfamiliar manuscript, we cannot simply assume we understand what its symbols are doing.
We have to ask:
What is actually legible?
Which relationships recur?
What can be inferred?
What remains unknown?
Apply that discipline to teaching.
A teacher writes:
Discuss the effectiveness of the writer’s use of language.
Experts may find the instruction perfectly normal.
But ask the learner:
What does discuss mean here?
What counts as effectiveness?
Which features count as use of language?
What is the expected answer shape?
How much evidence?
Does “effectiveness” require judgement?
Can the student actually decode the interface?
This is a powerful test:
Never confuse teacher familiarity with learner intelligibility.
The Problem Is Not Necessarily Bad Teaching
This distinction matters.
The issue is not:
teachers do not know English.
Nor:
schools teach English incorrectly.
Many structural pressures make compression inevitable.
Curricula contain large amounts of material.
Lessons have time limits.
Classes contain different learners.
Examinations need standardisation.
Teachers need efficient terminology.
Students progressively encounter specialised academic language.
So terms such as:
infer,
elaborate,
evaluate,
synthesise,
analyse
are useful.
The problem begins when the label is treated as though saying the label automatically installs the capability.
It does not.
Name ≠ capability.
Knowing the word analyse does not mean knowing how to analyse.
Knowing the definition of inference does not mean reliably making one.
Knowing PEEL does not guarantee coherent reasoning.
Knowing a model essay does not generate another model essay.
English Is Also the Student’s Knowledge-Access Layer
Reading introduces another dimension.
A text does not transfer knowledge directly into the mind.
The learner reconstructs meaning.
Vocabulary matters.
Syntax matters.
Background knowledge matters.
Inference matters.
Attention matters.
Relevant relationships have to be built across sentences.
Recent research continues to show that reading comprehension cannot be reduced to decoding words; vocabulary, topic knowledge and inference-making contribute to how successfully readers construct meaning from text.
That means weak comprehension reduces more than English marks.
It can reduce access to:
textbooks,
instructions,
articles,
research,
worked examples,
documentation,
feedback,
and independent study.
English becomes partly a knowledge-access layer.
Vocabulary Determines Resolution
Suppose a student knows only:
good,
bad,
angry,
happy,
problem,
thing.
The student can communicate.
But conceptual resolution is limited.
Now add:
beneficial,
efficient,
ethical,
sustainable,
counterproductive,
resentful,
indignant,
ambivalent,
constraint,
trade-off,
consequence,
mechanism.
These are not merely fancier words.
They allow finer distinctions.
Vocabulary therefore does at least two jobs.
It helps us communicate distinctions.
But it can also help us hold and manipulate distinctions while thinking.
This is why vocabulary development can alter the sophistication of:
reading,
reasoning,
writing,
discussion,
and subject learning.
Longitudinal research has also found vocabulary knowledge to be strongly associated with later reading-comprehension development.
Grammar Carries Relationships
Grammar is often taught as correctness.
Subject-verb agreement.
Tense.
Articles.
Prepositions.
Punctuation.
Those matter.
But grammar also carries logic.
Compare:
The plant died because it received too little water.
The plant died although it received enough water.
The plant died before it received water.
The plant may die if it receives too little water.
Much of the vocabulary remains constant.
The relationship changes.
Grammar encodes:
cause,
condition,
sequence,
contrast,
probability,
agency,
time,
scope.
So grammar failure is not merely cosmetic.
Sometimes it alters the model being communicated.
English Is Also an Output Interface
Even if the student understands perfectly, school usually requires the understanding to become observable.
The learner must:
say it,
write it,
draw it,
calculate it,
select it,
or otherwise demonstrate it.
For many subjects, language is part of this output.
That creates another distinction:
knowledge possessed ≠ knowledge successfully expressed.
A learner might possess a correct concept yet produce an ambiguous answer.
Another may understand a passage but write something too vague for the examiner to recognise.
Another may verbally explain an excellent argument but fail to organise it on paper.
This is a translation failure.
The thought exists.
The output does not preserve it.
English Is Also the Feedback Interface
Now consider what happens after the work is marked.
The teacher writes:
Develop your explanation.
Be more precise.
Link this to the question.
Avoid generalisation.
Analyse the evidence.
Improve coherence.
Those comments are new commands.
The learner must decode them too.
If “be more precise” is not converted into a specific executable operation, the feedback remains inert.
The student sees it.
Maybe highlights it.
Maybe nods.
Then writes the same way next week.
Feedback only becomes educationally useful when the learner can turn it into action.
feedback language
→ interpreted failure
→ repair operation
→ changed attempt.
Again, English sits in the control loop.
English Eventually Becomes the Student’s Internal Instruction System
This may be one of the deepest functions of language in education.
At first, the teacher says:
Read the question again.
Later, the student tells themselves:
Read the question again.
The teacher says:
What evidence supports that?
Eventually:
What evidence supports this?
The teacher says:
You have not explained the relationship.
Eventually the learner notices:
I have made a claim but not explained why it is true.
External instruction becomes partially internalised.
The student begins using language to direct personal cognition.
check;
compare;
reconsider;
slow down;
retrieve;
verify;
organise;
revise.
English therefore does not only transmit commands between people.
It can become part of the learner’s self-command architecture.
Weak English Can Increase Cognitive Load
Imagine trying to solve Mathematics while simultaneously struggling to understand the wording.
Part of the learner’s limited working capacity is now occupied by language decoding.
The mathematical operation still has to happen.
So do memory retrieval and error monitoring.
Similarly, writing requires idea generation, vocabulary retrieval, sentence construction, organisation and monitoring to compete for limited cognitive resources.
Research consistently links working-memory processes with reading and learning, including the ability to follow instructions and hold information while processing it.
This gives us another mechanism:
difficult language interface
→ greater processing demand
→ fewer resources available downstream.
Again, this will not explain every failure.
But it can explain why a learner performs much better once the wording of a problem has been clarified.
So What Is the Main Problem With Teaching English?
We can now state it more precisely.
We often teach English as a collection of outputs while education depends on English as an operating interface.
Students are taught:
grammar,
vocabulary,
comprehension,
composition,
oral communication.
All important.
But underneath them is another system:
receive
→ parse
→ interpret
→ model
→ connect
→ reason
→ select
→ express
→ monitor
→ repair.
This is what makes English transferable across the curriculum.
The Problem With Teaching Only the Outputs
Consider the normal visible outputs.
A correct answer.
A good essay.
A vocabulary list.
A comprehension worksheet.
A grammar exercise.
A model oral response.
These show what successful performance can look like.
But Alien 2 asks:
What hidden capability generated this output?
That question changes tuition.
For a good composition:
What generated the idea?
How was relevance decided?
How was the structure chosen?
How were examples retrieved?
How was vocabulary selected?
How was coherence monitored?
How was weak material removed?
For a comprehension answer:
How was the question interpreted?
How was evidence located?
How was the inference generated?
How were alternatives eliminated?
How was the final answer calibrated?
Now tuition begins teaching the production architecture rather than only the product.
This Is Where the English Mind Field Connects
Our English Mind Field work fits directly here.
When the student receives a task, the mind does not retrieve a complete polished answer from storage.
A field of possibilities activates.
Words.
Knowledge.
Memories.
Examples.
Relationships.
Interpretations.
Potential directions.
The student must navigate:
activate
→ map
→ connect
→ constrain
→ select
→ express.
A weak English system can interfere at each stage.
The student may not understand the prompt.
May lack the vocabulary needed to activate knowledge.
May misread relationships.
May generate ideas but fail to organise them.
May construct a good internal model but fail to encode it accurately.
Therefore:
“My child has no ideas”
may actually mean:
“My child cannot reliably navigate from linguistic input to usable output.”
That is a very different diagnosis.
The Earlier Weak Link Matters
Suppose a student writes poor Science explanations.
The visible failure is:
weak written answer.
Possible earlier causes include:
scientific knowledge missing;
command word misunderstood;
causal relationship unclear;
vocabulary insufficient;
grammar obscures causality;
answer structure unknown;
idea exists but cannot be translated;
time pressure collapses expression.
These are not interchangeable.
This is why our wider Learning Continuity work distinguishes different failure types.
A Missing Node:
required knowledge is absent.
A Broken Edge:
knowledge exists but relationships do not.
A Weak Link:
the connection works only under easy conditions.
A Wrong Edge:
an incorrect relationship has been learned.
A Routing Failure:
the learner has the knowledge but cannot retrieve it when needed.
A Translation Failure:
the internal answer exists but does not survive into language.
A Transfer Failure:
the student can perform in one familiar context but not another.
A Calibration Failure:
the learner cannot judge whether the answer is sufficient.
A Regulation Failure:
attention, time or emotional load disrupts available capability.
Now “weak English” becomes diagnosable.
English Tuition Should Therefore Start Earlier Than the Essay
If a student struggles with composition, it is tempting to begin at composition.
But the earliest weak link might be:
reading.
Or vocabulary.
Or knowledge.
Or sentence control.
Or idea relationships.
Or task interpretation.
Or retrieval.
Good English tuition therefore moves backwards when necessary.
visible failure
→ trace dependency
→ locate earlier weakness
→ repair
→ reconnect downstream.
This is one reason endlessly assigning more essays can fail.
The essay is the final assembly surface.
The broken component may live much earlier in the chain.
English Tuition Also Needs to Move Forward
Repair alone is not enough.
A student who understands English during tuition but cannot use it elsewhere has not completed the learning cycle.
The capability must travel:
English lesson
→ school English;
English
→ Mathematics;
English
→ Science;
English
→ Humanities;
guided practice
→ unfamiliar task;
untimed work
→ examination pressure;
teacher prompting
→ independent control.
This is Learning Continuity.
The student does not merely learn something.
The learner remains able to reconnect to it when the environment changes.
The Centre and the Edge
Another idea from our recent civilisation research helps here.
The Centre prefers:
standard terminology,
efficiency,
measurable outcomes,
curriculum coverage,
exam formats.
These are necessary.
But the Edge is where unusual failures appear.
The student who knows the Mathematics but cannot read the problem.
The student who can explain orally but not write.
The child whose vocabulary is strong but inference is weak.
The student whose answers are excellent until time pressure arrives.
The learner who memorises every model essay but cannot generate an unfamiliar one.
A robust tuition system needs both:
Centre: teach the common architecture.
and:
Edge: detect the learner-specific exception.
The danger begins when the standard programme becomes so strong that we stop seeing the student who does not fit it.
Do Not Narrow Too Early
Another finding from the newer education research is relevant.
Students need both breadth and depth.
If English becomes only:
exam question
→ technique
→ answer template,
the corridor can become too narrow.
The student may improve at recognisable tasks while the wider language system remains underdeveloped.
A healthier trajectory is closer to:
broad language exposure
→ conceptual foundations
→ exploratory use
→ structured practice
→ increasingly demanding application
→ examination control
→ independent judgement.
The examination corridor matters.
But it should sit on top of a larger language field.
Otherwise we may produce students who know exactly what to do when the worksheet resembles training—and very little when it does not.
AI Makes This Problem More Important
Artificial intelligence can now generate:
paragraphs,
summaries,
vocabulary,
arguments,
examples,
explanations,
model answers.
That reduces the cost of producing educational output.
But it increases the importance of another capability:
Can the student inspect what has been produced?
Is the instruction understood?
Is the answer relevant?
Is the claim supported?
Is the explanation causal?
Has a qualification been lost?
Is the word being used correctly?
Does the output actually answer the task?
If generation becomes cheaper while judgement remains weak, students can produce more language without gaining more understanding.
So the future value of English tuition may move further towards:
comprehension,
mapping,
navigation,
questioning,
evidence,
calibration,
judgement,
self-correction.
The student needs to become capable not simply of generating language, but of governing language.
What English Tuition Should Actually Teach
This gives us a more complete architecture.
English tuition should strengthen the student’s ability to:
1. Decode
What do the words mean?
2. Parse
How are the words related?
3. Reconstruct
What situation, idea or task does the language describe?
4. Identify the Directive
What am I being asked to do?
5. Activate
What knowledge is relevant?
6. Map
How do the relevant ideas connect?
7. Constrain
Which possibilities fit the question?
8. Reason
What follows from the evidence and relationships?
9. Encode
How should the internal thought become spoken or written English?
10. Monitor
Did the meaning survive?
11. Repair
If it did not, where did the chain break?
12. Transfer
Can I do this when the wording, topic or subject changes?
That is far more than grammar tuition.
And far more than composition tuition.
It is closer to building a language operating system for learning.
A Better English Tuition Loop
The resulting eduKateSG loop becomes:
Receive
→ Decode
→ Parse
→ Model
→ Map
→ Select
→ Execute
→ Express
→ Check
→ Repair
→ Transfer.
Then repeat.
Every comprehension passage can train this.
Every composition can train this.
Every oral discussion can train this.
Even Mathematics and Science questions can reveal whether the language interface is functioning correctly.
A Student Should Eventually Be Able to Debug the Compiler
At first the tutor detects the error.
You misunderstood “contrast”.
Later the tutor asks:
What operation does “contrast” require?
Eventually the student sees the word and independently activates:
differences, organised on comparable dimensions.
At first:
You answered what happened, but the question asked why.
Eventually:
I have described instead of explained.
At first:
This sentence changes your intended meaning.
Eventually:
That connector gives the wrong causal relationship.
This progression matters.
Tutor debugging
→ shared debugging
→ learner debugging.
The student is acquiring metacognitive control.
The Goal Is Not Permanent Tuition
This connects to perhaps the deepest conclusion from the recent EducationOS work.
A successful educational system should eventually ask:
What can the learner carry without us?
Not merely:
What can the learner complete while we are present?
The destination is portability.
Knowledge the student can retrieve.
Methods the student can reconstruct.
Language the student can control.
Judgement the student can apply.
Mistakes the student can detect.
Learning the student can continue independently.
That is why teaching cannot stop at finished outputs.
The learner needs some access to the generator underneath them.
From Alien 1 to Alien 2
The educational journey can now be described very simply.
Alien 1
The learner encounters:
the formula;
the rule;
the essay;
the question;
the rubric;
the model answer;
the examination command.
And asks:
What does this thing mean?
That is decoding.
Alien 2
The learner goes deeper and asks:
What problem produced this?
What relationships make it work?
What capability generates it?
How could I reconstruct it myself?
That is generative understanding.
Strong tuition needs both.
Students must be able to read the finished civilisation they have inherited.
But eventually they must also become capable of rebuilding and extending parts of it.
Why This Matters Beyond English
Human civilisation stores enormous amounts of capability in language.
Books.
Laws.
Scientific papers.
Manuals.
Instructions.
Contracts.
Software documentation.
Medical guidance.
Historical records.
Educational materials.
Institutional procedures.
Language allows knowledge generated by one mind to affect another mind across space and time.
But this system has a condition:
the receiver must reconstruct enough of the intended meaning.
If the reconstruction fails, the information exists but the capability does not arrive.
This gives English education a larger purpose.
We are not merely teaching children to make fewer grammar mistakes.
We are strengthening part of the interface through which they access accumulated human knowledge.
The Main Danger
The danger is therefore not simply:
weak English marks.
It is silent distortion.
The student reads something.
Believes it has been understood.
Builds new knowledge on top of the misunderstanding.
That new knowledge becomes the foundation for another concept.
Then another.
Eventually the visible problem appears far downstream.
This resembles building software on a corrupted dependency.
The higher layers may look increasingly sophisticated.
The lower-layer interpretation remains wrong.
So when English is weak, aggressively unpacking more and more sophisticated Mathematics, Science, Humanities or AI-generated material can sometimes increase the amount of misunderstood information rather than increase capability.
More input is not automatically more learning.
The receiving system must be able to reconstruct it.
This Is Why English Tuition Sometimes Needs to Slow Down Before It Speeds Up
Parents naturally want improvement.
More vocabulary.
More papers.
More compositions.
More comprehension.
More techniques.
Sometimes that is exactly what the student needs.
Sometimes the faster route begins by slowing down.
What did that sentence mean?
What did the question ask?
Why does this connector matter?
What relationship does the grammar encode?
Which word caused the misunderstanding?
What did you think the teacher meant?
Can you explain the instruction in your own words?
Can you show me the task before solving it?
These apparently simple questions can reveal failures that hundreds of worksheets hide.
Frequently Asked Questions
Is English really necessary for Mathematics?
Mathematical capability is distinct from English ability.
However, where Mathematics is presented through English—particularly word problems, instructions and explanations—the learner must understand enough language to construct the intended mathematical problem.
Research on mathematical word problems supports the importance of text comprehension in successful problem representation and solution.
Can a student be good at Science but weak at English?
Certainly.
The student may understand scientific concepts while struggling to express them.
The reverse is also possible: fluent English does not automatically produce scientific understanding.
The important task is to distinguish concept, language input, reasoning and language output.
Should English tuition teach Mathematics and Science vocabulary?
Where language patterns interfere with learning, cross-domain examples can be extremely useful.
But the purpose is not to turn English tuition into Mathematics or Science tuition.
It is to teach students how language encodes:
conditions,
relationships,
commands,
causes,
comparison,
evidence,
scope,
and precision.
Why can a child understand when the tutor explains the question but not solve it independently?
The tutor may be performing part of the language decoding or task construction for the learner.
Once the question has been translated into a usable internal model, the student’s subject knowledge becomes available.
That is diagnostically useful.
The next objective is to teach the student to perform that translation independently.
Why doesn’t doing more worksheets always solve the problem?
Because worksheets primarily expose outputs.
If the recurring failure lives earlier in the chain—decoding, routing, inference, translation or calibration—additional volume can reproduce the same error without repairing it.
Are model essays bad?
No.
Model essays are valuable.
The danger is treating the model as though exposure to the finished product automatically transfers the capability that generated it.
A stronger lesson reverse-engineers the model:
Why this idea?
Why this order?
Why this evidence?
Why this word?
Why this qualification?
What alternatives were possible?
Then the student attempts a different task.
Is “Command and Directives Compiler” a scientific term?
No.
It is an eduKateSG systems metaphor.
Human language processing is not computer compilation.
We use the metaphor to highlight an educational fact that is easy to miss: students frequently have to transform linguistic instructions into an internal representation of what operation they are supposed to perform before subject knowledge can be used.
How English Tuition Works
So how should English tuition work?
Not:
worksheet
→ mark
→ next worksheet.
And not simply:
technique
→ memorise
→ examination.
A stronger system is:
observe the learner
→ inspect the language interface
→ identify the earliest weak link
→ reconstruct the hidden operation
→ practise it explicitly
→ reconnect it to the wider English system
→ test it in another context
→ compress it for fluent use
→ transfer control to the student.
Sometimes the repair is vocabulary.
Sometimes grammar.
Sometimes comprehension.
Sometimes background knowledge.
Sometimes idea mapping.
Sometimes writing.
Sometimes examination technique.
Sometimes the student knows all of those things individually but cannot route them correctly.
Good tuition finds out which.
English Is Not Just Another Subject
This may be the final point.
A school timetable can place English in one box.
Monday, 10.00 am.
English.
Then Mathematics.
Then Science.
Then Humanities.
But the student’s learning system does not respect those timetable borders so neatly.
English travels.
Into the Mathematics question.
Into the Science explanation.
Into the History source.
Into the Geography essay.
Into the teacher’s feedback.
Into the examination rubric.
Into the textbook.
Into Google.
Into artificial intelligence.
Into instructions.
Into conversation.
Eventually, into the student’s own internal voice.
That is why English weakness can sometimes have consequences far beyond the English paper.
And it is why English tuition cannot be understood merely as:
fixing grammar and improving composition.
We are strengthening an interface between:
instruction and action;
information and understanding;
knowledge and reasoning;
thought and expression;
teacher and learner;
learner and examination;
learner and the accumulated knowledge of civilisation.
Alien 1 reminds us that the student often encounters the finished compressed world.
Alien 2 reminds us that good education must sometimes reconstruct the world underneath it.
The English Mind Field shows us that ideas need routes.
Learning Continuity reminds us that capability must survive time and context.
The weak-link model tells us that the visible error may occur far downstream from its cause.
And the latest EducationOS work gives us the final test:
What can the learner eventually carry without us?
That may be the main problem with teaching English.
We have spent a great deal of time teaching students the outputs of language.
The next step is to teach them increasingly to control the system underneath the outputs.
Because when that system becomes strong, English stops being merely another subject the child studies.
It becomes one of the tools through which the child can study almost everything else.