How to perform in the new G2 SEC examinations improves faster when the learner repairs the right problem first. A wrong answer can come from several different causes: missing knowledge, failure to recognise the method, execution error, interpretation error, or whole-paper control. Treating all of them as “carelessness” or “weak topic knowledge” wastes revision time.
This forty-fifth Learner’s Guide develops an error-triage system for G2 English, Mathematics and Science. The central rule is: repair the earliest failure that made the final answer wrong. If the task was misread, grammar polishing comes too late. If the method was recognised correctly but a sign was lost, reteaching the whole topic is unnecessary. If the answer was correct but unfinished because time expired, the repair belongs to paper control rather than content.
For 2027, SEAB lists G2 English Language K200, Mathematics K210 and G2 Science combinations K223, K224 and K225 for school candidates. Use the official G2 syllabus directory for current subject documents. The triage categories below are eduKateSengkang teaching tools, not official score categories.
The five error families
- Knowledge gap — the concept, vocabulary, fact or method is missing.
- Recognition gap — the learner knows the method when prompted but does not identify it independently.
- Execution gap — the learner chooses the right method but performs it inaccurately.
- Interpretation gap — the learner obtains useful work but answers the wrong practical or linguistic question.
- Control gap — time, attention, checking or recovery causes otherwise accessible marks to be lost.
These families can overlap, but one usually appears first in the chain. That first failure is the highest-value repair target.
Why the first failure matters
Suppose a Mathematics answer is numerically wrong. If the learner chose the wrong percentage base, the problem is recognition or modelling. If the base was correct but 0.8 × 90 was entered incorrectly, the problem is execution. If the calculation was correct but the learner reported the discount instead of the final price, the problem is interpretation.
Three wrong answers can require three different interventions even when they occur in the same chapter.
The triage sequence
- What was the task?
- What did the learner think the task was?
- What method or response form was chosen?
- Where did the working first become invalid?
- Was the final answer interpreted and communicated correctly?
- Did timing or attention change the outcome?
This sequence moves from earliest to latest. Do not start with the final wrong number.
Knowledge gap
A true knowledge gap appears when the learner cannot state or reconstruct the relevant concept even after the task has been identified.
Examples include not knowing the meaning of a word in context, not knowing a required algebraic relationship, or not knowing the scientific mechanism needed for an explanation.
Signs of a knowledge gap
- blank response even after the task is clarified;
- random unrelated facts;
- method cannot be explained after a prompt;
- definition is absent or seriously confused;
- the learner depends on copying the model answer rather than reconstructing it.
The repair is teaching, not more timed papers.
Recognition gap
A recognition gap appears when the learner can solve the problem once the method is named but does not identify that method independently.
This is common in mixed Mathematics, English question types and Science command-word work.
Signs of a recognition gap
- correct immediately after the tutor says “use ratio” or “this is inference”;
- strong performance on labelled worksheets but weak performance in mixed sets;
- several methods tried randomly before the right one appears;
- chapter knowledge seems secure but unfamiliar wording causes collapse.
The repair is mixed practice and classification, not more identical examples.
Execution gap
An execution gap occurs after a valid method has already been selected.
Typical examples are sign errors, calculator-entry mistakes, agreement errors, missing units, copied values, grammar slips or incomplete scientific chains where the concept itself is known.
Signs of an execution gap
- setup is correct;
- the learner can explain the intended route;
- one local step breaks the solution;
- the error can often be fixed without reteaching the concept.
The repair is targeted practice at the fragile transition plus checking.
Interpretation gap
An interpretation gap occurs when the working is mathematically, linguistically or scientifically useful but the final answer does not satisfy the task.
Examples include giving 4.2 buses instead of the minimum whole number, writing an observation when Science asks for an explanation, or giving a correct English detail that does not answer the requested inference.
Signs of an interpretation gap
- correct intermediate result;
- wrong final label or unit;
- failure to return to context;
- response form mismatches the command word;
- answer contains true information but not the required information.
The repair is task-return and final-answer discipline.
Control gap
A control gap appears when the learner could probably perform the task under ordinary conditions but loses access because of time, fatigue, attention drift or poor skip-return decisions.
This does not mean every late-paper error is “just pressure”. The earlier categories must still be checked.
Signs of a control gap
- accuracy drops sharply late in the paper;
- easy questions become rushed after one difficult item;
- blank marks remain despite accessible later questions;
- correct answers are changed without evidence;
- checking time is spent on low-risk work while unfinished items remain.
The repair belongs to pacing, recovery, stamina and mark security.
Do not diagnose from one answer alone
One unusual error may not represent a stable weakness. Use several examples before declaring a pattern.
A learner can lose one sign by chance without having a general algebra problem. Triage should be evidence-based.
The three-example rule
Before building a new intervention around a suspected weakness, look for the same failure in at least a few different contexts where practical.
If the pattern repeats, the diagnosis becomes stronger. If it disappears, record the original error but avoid overcorrecting.
English triage: comprehension
Wrong comprehension answers can originate at several levels:
- knowledge — vocabulary or syntax prevents understanding;
- recognition — question type is misclassified;
- execution — paraphrase changes meaning;
- interpretation — evidence is correct but command is not answered;
- control — learner rushes and skips the relevant lines.
The same red cross on the page does not tell you which level failed.
English triage: writing
A weak paragraph may come from missing ideas, poor organisation, grammar execution, register mismatch or time pressure.
Before assigning more essays, identify whether the learner cannot generate, cannot structure, cannot express or cannot sustain the skill under time.
English triage: oral
A short oral answer may reflect limited ideas, slow retrieval, uncertainty about the question, weak development or anxiety after an earlier difficult prompt.
The next practice should match the cause: idea generation, response ladder, follow-up drills or recovery.
Mathematics triage: word problems
A wrong word problem can begin with misreading, modelling, method choice, algebra, arithmetic, unit conversion or final interpretation.
Write enough working to see the first wrong step. A bare final answer hides the diagnosis.
Mathematics triage: algebra
If the equation itself is wrong, reteach modelling. If the equation is right but transformation fails, practise the algebraic transition. If the solution is right but context rejects it, practise domain and feasibility.
Do not reteach the entire chapter when the error is local.
Mathematics triage: graphs
Separate axis reading, scale, plotting, gradient, interpretation and context.
A learner can understand graphs generally yet fail one layer repeatedly.
Science triage: structured response
A wrong Science answer may reflect missing content, wrong discipline mode, command-word mismatch, evidence selection, mechanism or claim strength.
The repair should begin at the earliest failed layer.
Science triage: experiments
Experimental questions contain variables, apparatus, measurement, data, reliability, limitation and improvement.
A generic “repeat more times” answer may indicate recognition of reliability language without understanding the actual weakness.
Worked triage chain one: English inference
Question asks why a speaker hesitated before agreeing. The learner copies a sentence showing the hesitation but never explains the reason.
Knowledge may be adequate because the relevant sentence was located. Recognition is partly adequate because the learner knows where to look. The first failure is interpretation: evidence was supplied where explanation was required.
Repair: practise evidence → meaning pairs, not more vocabulary lists.
Worked triage chain two: English paraphrase
The learner identifies the correct source point but rewrites “may improve access” as “will improve access”.
Task recognition is correct. Evidence selection is correct. The first failure is execution of paraphrase fidelity. Repair certainty words such as may, might, likely and will.
Use Vol 0042 for the detailed fidelity system.
Worked triage chain three: Mathematics percentage
The learner sees a sale price after discount and calculates 20% of the sale price to recover the original.
Arithmetic may be flawless. The first failure is modelling the percentage base. More calculator drills will not help.
Repair: represent original = 100%, sale = remaining percentage, then form the equation before calculating.
Worked triage chain four: Mathematics sign
The learner writes the correct equation, expands correctly, then loses a negative sign while rearranging.
Knowledge and recognition are secure. The failure is local execution. Repair sign transitions and use substitution to check the final value.
Do not return to the beginning of the chapter unless the sign mistake reveals a deeper misunderstanding.
Worked triage chain five: Science command word
The learner accurately describes a graph when the question asks to explain the trend.
Content may be present. Evidence use is present. The first failure is operation recognition. Repair the difference between describe and explain.
Use Vol 0024.
Worked triage chain six: Science evidence
The learner states a correct mechanism but ignores the data pattern supplied.
Knowledge is strong; evidence integration is weak. The repair is not memorising the mechanism again. It is evidence → concept → mechanism practice.
Worked triage chain seven: time trap
A learner spends eight minutes on one low-value Mathematics item, then rushes three accessible questions.
The original difficult item may contain a knowledge or recognition gap, but the broader score loss is control. The repair must include a stop condition and return cue.
One local weakness should not be allowed to create several new errors elsewhere.
Worked triage chain eight: late-paper English
The learner writes strong early paragraphs but the final paragraph becomes repetitive and grammatically unstable.
If the same paragraph is strong when written fresh, the dominant problem is stamina or time control rather than writing knowledge.
Repair with timed paragraph sequencing and final-third practice.
The repair menu
- Knowledge gap → reteach, retrieve, spaced review.
- Recognition gap → mixed practice, contrast pairs, method classification.
- Execution gap → targeted micro-drills, check at fragile transition.
- Interpretation gap → return-to-task routines, final-answer sentences, feasibility checks.
- Control gap → timed segments, skip-return rules, stamina, mark security.
The repair should match the diagnosed family. A mismatch wastes effort.
Do not prescribe full papers for every error
Full papers are integration tests. They are inefficient tools for repairing one narrow sign error, one pronoun problem or one command-word confusion.
Use full papers after targeted repair to see whether the corrected skill transfers back into the larger system.
Do not prescribe micro-drills forever
The opposite mistake is keeping the learner in isolated drills after the skill has stabilised.
Once a repair works, place it back into mixed and timed contexts. A skill is useful only if it survives integration.
The repair ladder
- Explain the error.
- Redo the original without copying.
- Attempt a near-transfer question.
- Attempt a mixed question where the method is hidden.
- Attempt under realistic time.
- Retest after a delay.
This ladder turns correction into durable change rather than answer copying.
The first retest
The first retest should be similar enough to isolate whether the repair worked.
If the learner is repairing percentage base selection, keep the arithmetic simple. Do not add a complicated graph and several constraints at the same time.
The second retest
The second retest should change surface context while preserving the same deep skill.
This tests transfer. If the learner succeeds only when the question looks almost identical, the repair is not yet stable.
The delayed retest
Return after one or more days without reviewing first.
If the same error returns, the learning has not yet become durable. Use retrieval and spaced practice rather than immediate repetition alone.
The mixed retest
Place the repaired skill beside similar but different tasks.
Recognition gaps often appear only when the learner must choose among methods rather than execute a named method.
The timed retest
Only after accuracy is reasonable should time become a major condition.
The aim is to see whether the repaired process survives time, not whether the learner can rush through it.
The repair stop rule
Stop targeted repair when the learner can demonstrate the skill independently across several contexts and after a delay.
Keep it in maintenance through ordinary mixed practice rather than drilling it indefinitely.
The repair escalation rule
Escalate to deeper reteaching when the same error persists despite targeted correction and the learner cannot explain the underlying relationship.
Repeated failure under multiple representations is stronger evidence of a concept gap than one wrong answer.
The error ledger should record causes
A useful ledger does not merely list “Question 7 wrong”. It records the first failure: missed condition, wrong method, sign error, unsupported conclusion, ambiguous pronoun or time trap.
Over several papers, cause patterns become visible.
The triage matrix
- Frequent + high-impact → repair first.
- Frequent + low-impact → automate with a small routine.
- Rare + high-impact → build a specific check.
- Rare + low-impact → record, but do not overreact.
This prevents revision from being driven by the most recent dramatic mistake rather than the highest-value pattern.
High-impact English errors
- misreading task;
- missing required content point;
- source meaning changed;
- paragraph drift;
- tone mismatch in a key communicative task.
These deserve priority over replacing a correct ordinary word with a more sophisticated one.
High-impact Mathematics errors
- wrong model;
- wrong method selection;
- constraint omitted;
- unit conversion corrupts a long chain;
- time trap leaves multiple later questions unfinished.
A single structural error can affect several marks.
High-impact Science errors
- wrong command operation;
- wrong discipline model;
- ignoring supplied evidence;
- mechanism missing from explanation;
- conclusion claims more than experiment supports.
Again, repair structure before vocabulary polish.
Triage after a mock paper
Review the paper twice. First, mark correctness normally. Second, ignore the score and identify the first failure in each meaningful error chain.
The second review produces the learning plan.
Triage after homework
Homework is useful because time pressure may be lower. Errors that remain even without timing are more likely to reflect knowledge, recognition or execution than whole-paper control.
Compare homework and timed-paper patterns before deciding the cause.
Triage after tutoring
If the learner succeeds only with prompts, record what prompt was needed.
“Use Pythagoras” indicates recognition was supplied externally. “Check your arithmetic” leaves recognition independent and supports an execution repair. Prompt type is diagnostic evidence.
Parents: ask what changed
After correction, ask “What will you do differently on the next question?”
A useful answer names a process: mark the constraint, write the equation, check the unit, compare both sources. “Be more careful” is too vague to guide future action.
Tutors: delay the rescue
Allow a short window for the learner to identify the failure before giving the method.
If the tutor always supplies the classification, recognition cannot become independent.
The triage worksheet
- Question / task.
- First wrong decision.
- Error family.
- Evidence for diagnosis.
- Repair activity.
- Fresh retest.
- Delayed retest result.
This is enough structure for a high-value review. Avoid turning the record into administrative work that takes longer than the learning.
The four-week triage cycle
Week 1 — classify
Use recent work to identify recurring error families.
Week 2 — repair
Target the highest-frequency or highest-impact causes with short drills.
Week 3 — transfer
Use mixed and unfamiliar contexts.
Week 4 — integrate
Use timed sections and one larger mock to see whether the repair survives whole-paper conditions.
Use exam-readiness thresholds
Use Vol 0033 to decide whether the repaired skill is still emerging, stable or exam-ready.
Use self-coaching for control errors
Use Vol 0041 when the learner knows what to do but needs a compact cue at the point of failure.
Use the subject hubs for concept repair
For genuine knowledge gaps, return to the Secondary English Learning Hub, Complete Mathematics Index or Complete Science Index.
Error triage should route the learner into the right teaching, not replace that teaching.
The PSLE bridge
The earlier habit Read Before You Solve already prevents one major error family: misreading.
At G2, triage extends the same discipline across modelling, execution, interpretation and paper control.
Use Examination Craft
For skip-return, pacing and checking systems, continue through the Examination Craft hub.
Final rule
Do not repair the last visible error first. Repair the earliest failure that made the answer go wrong.
Teach missing knowledge. Train recognition when the method is hidden. Drill execution when the method is known. Return to context when interpretation fails. Repair pacing when the paper—not the concept—caused the loss. Accurate diagnosis turns revision from more work into better work.
The error-triage decision tree
A practical triage decision tree begins with one question: Did the learner know what the question was asking? If not, repair task reading first. If yes, ask whether the learner knew what method or response form to use. If not, repair recognition. If the method was correct, inspect execution. If execution was correct, inspect interpretation and final communication. Only after these are secure should you attribute the loss mainly to whole-paper control.
This order matters because later errors can be symptoms of earlier ones. A student may look rushed because they spent too long searching for a method they could not recognise. The visible time problem is real, but the first cause is recognition.
Triage example: one wrong English answer, three possible diagnoses
Question: Why did the writer postpone the decision? Learner A copies the line describing the postponement but gives no reason. This is task interpretation. Learner B understands the question but does not know the meaning of “reluctant” in the relevant sentence. This is knowledge. Learner C knows both but writes a correct reason in such vague language that the pronoun has no clear referent. This is execution or communication.
All three need different practice despite receiving the same mark outcome.
Triage example: one wrong percentage answer
Learner A thinks a 20% discount means subtract 20 from the price. This is knowledge or concept. Learner B knows percentages but fails to recognise that the given sale price is 80% of the original. This is recognition. Learner C forms 0.8P = 72 correctly but keys 72 ÷ 0.08. This is execution. Learner D gets $90 but reports $18 because the question asked for original price and the learner stops at the discount. This is interpretation.
The red cross on the final line hides four different learning needs.
Triage example: one weak Science explanation
Learner A does not know the particle model. Knowledge gap. Learner B knows it but thinks the command “describe” requires a mechanism. Recognition gap. Learner C selects the mechanism but reverses cause and effect. Execution gap. Learner D explains correctly but ignores the data range and claims the relationship always holds. Interpretation and evidence-boundary gap.
Again, the visible answer is not enough; trace the reasoning.
The earliest-error rule
During review, identify the earliest point where a reasonable next step became impossible.
If the learner misread the question, everything after that may be irrelevant. If the question was read correctly but the wrong method was selected, later arithmetic should not distract from method selection. If the method is sound, then inspect execution.
The downstream-error rule
Do not count every consequence of one early error as a separate weakness.
One wrong scale factor can create five wrong later values. The repair target is scale factor selection, not five different arithmetic deficits.
The independent-error rule
If later subparts are independent and contain separate errors, classify them separately.
Triage should neither multiply one cause into many nor collapse several unrelated causes into one.
The high-frequency triage
Errors recurring across many tasks deserve attention even when each one costs only one mark.
Examples include missing units, ambiguous pronouns, wrong denominator, command-word mismatch and copied values. Small recurring losses can exceed the effect of one advanced topic weakness.
The high-impact triage
Some rare errors deserve priority because they corrupt large sections of a paper.
Wrong task interpretation in an extended English response, wrong model in a multi-step Mathematics question or wrong discipline mode in a Science structured item can affect several marks at once.
The controllable-first principle
When two weaknesses are equally important, begin with the one that can be changed most directly.
A learner may not eliminate all fatigue quickly, but can immediately install a final-unit check. Small reliable improvements build a stronger paper while deeper issues are being addressed.
The dependency principle
Repair prerequisite skills before downstream skills.
If paragraph ideas are missing, polishing transitions will not solve the main problem. If a Mathematics concept is absent, speed drills will only automate confusion. If Science evidence cannot be read, mechanism practice cannot yet connect to the question.
The evidence-for-diagnosis rule
- one error → hypothesis;
- repeated similar errors → stronger diagnosis;
- same error across contexts → likely underlying process issue;
- error disappears after one explanation → possibly local misunderstanding;
- error reappears after delay → learning not yet stable.
Use this hierarchy to avoid overdiagnosing from one bad day.
Triage and confidence
A learner may feel confident about a wrong method or uncertain about a correct one. Confidence is useful context but not diagnosis.
Use the written work, explanation and retest to decide the error family.
Triage and blank answers
A blank can mean missing knowledge, time expiry, inability to recognise the method, fear of committing, or a strategic skip awaiting return.
Ask what happened before treating a blank as evidence of ignorance.
Triage and changed answers
If a correct answer became wrong during checking, the issue may be answer-change threshold rather than content.
Use Vol 0029.
Triage and late-paper collapse
Compare the same skill early and late in the paper. If accuracy is strong early and weak late, stamina or control becomes more plausible.
If the skill fails everywhere, the cause is more likely knowledge, recognition or execution.
Triage and prompt dependence
Record the smallest prompt that unlocks the answer.
- “What is the command word?” → task recognition issue.
- “Which formula relates these quantities?” → method recognition issue.
- “Check your sign.” → execution issue.
- “What does this number mean?” → interpretation issue.
Prompt type tells you where independence breaks.
The two-day triage retest
Day 1: correct the error and explain the cause. Day 2: attempt a fresh task with no reminder.
If the learner independently avoids the error, the repair may be taking hold. If the same cue is still needed, the weakness remains emerging.
The one-week triage retest
Return later in a mixed set. This checks whether the repair survives delay and recognition demand.
A corrected answer remembered from yesterday is weaker evidence than a new correct answer produced after a week.
The triage-to-plan conversion
Every meaningful diagnosis should end in one concrete next action.
- Knowledge → relearn one concept and retrieve it tomorrow.
- Recognition → mix three similar question types and classify before solving.
- Execution → practise the fragile step five times with checking.
- Interpretation → write final-answer sentences linked to the task.
- Control → use a timed set with a stop condition and return mark.
If the plan cannot be stated this specifically, the diagnosis may still be too vague.
The no-more-work-without-purpose rule
Do not assign another worksheet simply because an answer was wrong.
First decide what the worksheet is meant to change. More volume without a target can reinforce the same process.
The repair-success question
After practice, ask: what behaviour is now different?
The learner may now mark the percentage base, reread the command word, bracket negative values, label evidence or move on after a time limit. Behaviour change is stronger evidence than familiarity with the correction.
The triage ceiling
Not every tiny error deserves a formal intervention. Some isolated slips should simply be corrected and monitored.
Triage is valuable because it prevents both underreaction and overreaction.
The final triage checklist
- What was the first wrong decision?
- Which error family best describes it?
- Is there repeated evidence for this diagnosis?
- What is the smallest repair that targets it?
- What fresh task will test the repair?
- When will it be retested after delay?
Six questions are enough to turn a wrong answer into a learning plan.
Final perspective
The most useful error is the one whose cause can be identified accurately.
A learner does not improve because mistakes are counted; improvement comes when the reasoning chain is inspected, the earliest failure is repaired and the new process survives a fresh task. Triage turns examination review from a list of lost marks into a map of what to teach next.
The final triage calibration
After several weeks of review, compare the distribution of error families rather than only the total score. A learner who moves from knowledge gaps to occasional execution slips has improved even if one difficult paper keeps the raw percentage similar. The nature of the errors has changed.
Likewise, a learner who knows more content but continues to lose marks through task misreading or poor time control has not yet converted that knowledge into examination performance. Triage makes this difference visible.
The repair priority should therefore be dynamic. Once one family stabilises, reduce its dedicated practice and redirect time to the next limiting factor. Revision becomes a moving queue rather than a permanent list of weaknesses.
The final standard is not zero errors. It is a learner who can identify why an error happened, apply the right repair and demonstrate a changed process on a fresh task without depending on the original correction.
Triage should improve the next decision
The value of diagnosis appears on the next question. If the learner identified a recognition gap, the next mixed task should begin with independent method selection. If the issue was execution, the next task should preserve the same method but place attention on the fragile step. If the issue was control, the next practice should include the timing condition that caused the failure.
This keeps review from becoming retrospective commentary. A useful diagnosis produces a changed practice condition, and that changed condition produces new evidence about whether the repair worked.
The final triage habit is therefore prospective: identify the earliest failure, design the smallest repair, and decide what fresh task will prove that the learner can now proceed differently.
The final triage discipline is to separate cause from symptom. A blank answer may be the symptom of missing knowledge, failed recognition or lost time. A wrong final number may be the symptom of a modelling error or a local arithmetic slip. Repair becomes efficient only when the underlying cause, not the visible symptom, determines the next practice task.
Once the cause is clear, the next practice should be small enough to test that cause directly before returning to full-paper conditions.
