JC 2 General Paper GP Tuition | How to Use Feedback and Marked Scripts

Marked scripts are not the end of a learning cycle. They are the beginning of the next one.

A JC2 student can receive a page covered in comments, corrections, ticks, question marks and underlining, understand every comment in the moment, and still make almost the same mistakes in the next essay or comprehension paper. The problem is not always lack of effort. It is often a transfer problem: feedback was read, but it was never converted into a new decision rule that survives the next unseen task.

This guide is about that conversion. It shows how to turn teacher feedback, tutor comments, consultation notes, marked scripts, model comparisons and self-review into better future General Paper performance. The aim is not to produce a perfect correction of yesterday’s paper. The aim is to make tomorrow’s unfamiliar paper less likely to trigger the same failure.

If you are looking for the wider JC2 route, begin at the JC2 General Paper tuition hub. For full paper mechanics, use the separate Paper 1 examination-performance guide and Paper 2 examination-performance guide. This page owns the feedback-to-transfer problem that sits across both papers.

The central problem: feedback is information, not improvement

Feedback can tell you that something went wrong. Improvement requires you to discover what decision produced the error, what alternative decision should replace it, and whether that replacement still works when the surface features of the task change.

That distinction matters because many corrections are script-specific. “This example is weak” may be true, but it does not yet tell you how to recognise a weak example next time. “You have not answered the question” identifies a serious failure, but not whether the cause was misreading the command word, narrowing the scope too early, drifting from the thesis, or answering a familiar topic instead of the actual proposition. “Needs evaluation” may point to missing judgement, shallow comparison, unsupported qualification, or a paragraph that never returned to the standard established by the question.

A useful feedback system therefore asks three questions every time: What happened? Why did it happen? What observable action will I take next time?

A 90-second diagnostic router

• If your comments keep repeating across different scripts, start with recurrence tracking.
• If you understand comments only after a teacher explains them, start with feedback decoding.
• If your corrections look excellent but new work stays unchanged, start with unseen transfer testing.
• If your essay feedback is dominated by relevance, argument, examples or evaluation, use the Paper 1 transfer system.
• If your comprehension feedback is dominated by interpretation, precision, summary or application, use the Paper 2 transfer system.
• If you receive too many comments to act on, use the priority ladder.
• If feedback from different teachers appears to conflict, use the conflict-resolution protocol.
• If your main problem is timing, panic or unfinished work, connect this page to the A-Level GP exam-day strategy.

The seven-stage feedback transfer loop

1. Capture: collect the exact comment, mark loss, annotation or error pattern.
2. Classify: decide what kind of problem it is.
3. Explain: identify the likely cause in your thinking or execution.
4. Repair: make a small, explicit change to the failed decision or process.
5. Rehearse: practise the repaired decision on a nearby example.
6. Transfer: test it on an unseen task where the wording or topic changes.
7. Track: record whether the error disappeared, weakened, moved or returned.

The loop matters because a correction without transfer can create an illusion of learning. You can rewrite an essay after seeing comments and produce a much stronger version because the original problem, evidence and direction are now familiar. That proves you can improve the old script with help. It does not yet prove that your future question selection, planning, paragraph decisions or checking behaviour has changed.

Step 1: Capture the feedback before interpreting it

Students often jump immediately from a comment to a conclusion. “Vague” becomes “my English is bad.” “Weak evidence” becomes “I need more current affairs.” “Not answering the question” becomes “I need to write more carefully.” These translations are too broad to be useful. Start by preserving the evidence.

• Copy the exact comment or annotation.
• Record where it occurred: question choice, introduction, body paragraph, conclusion, short-answer item, summary, application response, or final checking.
• Record the consequence if known: mark loss, unclear reasoning, incomplete answer, repeated language error, time loss or unfinished section.
• Keep the surrounding sentence or paragraph so the comment does not lose context.
• If the feedback was spoken in a consultation, write the teacher’s key instruction in your own words immediately afterwards and check that you have not distorted it.

Capture is deliberately boring. That is a strength. It prevents emotional summaries from replacing evidence.

Step 2: Classify the error family

A long marked script becomes manageable when comments are grouped into recurring families. The exact labels can be adapted, but a useful JC2 system normally needs the following categories.

Question interpretation and scope

Did you answer the task that was actually set? Common failures include ignoring a qualifier, treating an absolute claim as a general topic, discussing only one part of a comparison, or substituting a rehearsed issue for the precise proposition.

Thesis and line of argument

Did the essay establish a defensible position and sustain it? A thesis can fail because it is merely descriptive, because its criteria are unclear, because later paragraphs do not support it, or because the conclusion quietly changes the position.

Reasoning and paragraph architecture

Did each paragraph move from claim to explanation, evidence, qualification and relevance? Many students have enough content but do not make the causal or comparative link explicit.

Evidence and examples

Was the example accurate enough, specific enough and actually connected to the claim? More examples are not automatically better. One well-chosen example whose mechanism is explained may outperform several name-drops.

Evaluation and judgement

Did the essay compare significance, conditions, trade-offs, time horizons or affected groups? Evaluation is not a decorative “however.” It is the work of deciding what matters under the terms of the question.

Language precision

Did wording express the intended meaning accurately? Track recurring grammar, sentence-boundary, reference, vocabulary, register and syntax problems separately from idea problems.

Paper 2 interpretation

Did you identify what the passage actually meant, rather than what you expected it to mean? This includes reference tracking, paraphrase, inference, tone, relationship between claims and answer-scope control.

Summary control

Did you select the right points, compress them accurately, avoid duplication and stay within the task’s boundaries? Summary failure can come from selection, paraphrase, compression or counting rather than general reading weakness.

Application and synthesis

Did you connect passage ideas to a relevant context with specific reasoning? A response can mention local or personal examples yet remain weak if it does not explain how the passage claim applies, fails, changes or requires qualification.

Execution and time

Was the knowledge present but inaccessible under pressure? Track late starts, overlong plans, slow reading, excessive rewriting, poor checking order and failure to abandon low-return work.

Step 3: Separate symptom from cause

This is the step that turns feedback into diagnosis. A comment describes the visible output. The cause often happened earlier.

• Symptom: “Example is not relevant.” Possible causes: question misunderstood; claim too broad; example chosen by topic similarity rather than mechanism; paragraph changed direction after the example was selected.
• Symptom: “Explanation too brief.” Possible causes: student assumed the link was obvious; knowledge was memorised as facts without mechanism; time pressure caused compression; the claim itself was too vague to explain.
• Symptom: “Answer lacks precision.” Possible causes: passage reference not tracked; key distinction collapsed; paraphrase replaced a narrow term with a broader one; student answered from memory rather than returning to the text.
• Symptom: “Needs evaluation.” Possible causes: no criteria in thesis; counterargument treated as a separate paragraph instead of evidence for judgement; student knows both sides but not how to compare them; conclusion delayed all judgement until the end.
• Symptom: “Incomplete paper.” Possible causes: planning too long; rereading without purpose; perfectionism; weak time checkpoints; difficulty leaving one uncertain answer and returning later.

Do not accept the first explanation automatically. Treat it as a hypothesis. The fastest way to test a cause is to design a small task where only that suspected weakness is stressed.

Step 4: Write a replacement decision rule

“Be more careful” is not a repair. “Use better examples” is not a repair. “Evaluate more” is not a repair. A repair must be observable enough that you can tell whether you did it.

Useful decision rules often take an if–then–check form:

• If a question contains a qualifier such as “most,” “only,” “too much” or “in today’s world,” then underline the qualifier and make every planned paragraph answer it, not just the broad topic.
• If I use an example, then write one sentence explaining the mechanism that makes it evidence for the claim before moving on.
• If a paragraph presents a counterargument, then state what condition would make it stronger or weaker and return to the essay’s criterion.
• If a short-answer response uses a broad synonym, then compare it with the passage wording and ask whether any meaning has been lost or expanded.
• If I cannot decide between two interpretations, then locate the smallest phrase in the passage that would make one interpretation impossible.
• If I reach a time checkpoint behind schedule, then reduce low-value elaboration before sacrificing a whole later section.

A good replacement rule is small enough to remember under pressure and specific enough to change behaviour.

Step 5: Rehearse close to the original error

After a correction, practise on a nearby task before jumping to a full paper. If the problem was example relevance, take five claims and choose or reject examples. If the problem was question scope, classify ten prompts by command and qualifier. If the problem was summary duplication, compress a short passage twice and compare overlapping points. If the problem was evaluation, take six claims and write one conditional judgement for each.

This short rehearsal isolates the repaired decision. It gives you more repetitions than a full paper and makes failure easier to interpret.

Step 6: Transfer to an unseen task

Transfer is the decisive test. Change the topic, wording, passage, context or time pressure. Do not look at the old correction while attempting the new task. The question is whether the new rule appears spontaneously when the surface features change.

• A repaired essay-introduction problem should be tested on an unfamiliar question family.
• A repaired evidence problem should be tested with a different domain, not the same favourite example.
• A repaired comprehension paraphrase problem should be tested with a new passage containing different vocabulary but the same precision challenge.
• A repaired timing problem should be tested under a clock, not only in untimed review.
• A repaired application problem should be tested using a context you did not discuss during the consultation.

If the error returns, that is not wasted work. It means the repair has not yet become portable.

Step 7: Track recurrence, not feelings

Students often judge improvement by how confident a lesson felt. A better measure is recurrence. For each important error family, track whether it appears across the next three to five unseen attempts.

• 0: not observed.
• 1: appeared once but self-corrected.
• 2: appeared and required a prompt to fix.
• 3: repeated strongly enough to affect marks or completion.

The score is not a grade. It is a recurrence signal. A problem that moves from 3 to 1 is changing even if the total mark has not yet risen because another weakness has become the new bottleneck.

Build a feedback ledger

A feedback ledger is a compact record of high-value recurring problems. It should not become a second textbook. One row can contain:

• date and task;
• exact feedback;
• error family;
• suspected cause;
• replacement rule;
• repair drill;
• unseen transfer task;
• recurrence result;
• next action.

The ledger becomes especially useful late in JC2 because it compresses months of marked work into a small set of live constraints. Instead of revising everything equally, you can see which errors still survive pressure.

The priority ladder: not all feedback deserves equal time

A heavily annotated script can produce twenty possible actions. Trying to repair all twenty at once usually creates shallow attention. Prioritise by expected return.

1. Task failure: misunderstanding the question, leaving sections incomplete, violating answer scope.
2. Structural failure: unstable thesis, broken reasoning chain, poor passage interpretation, repeated summary selection errors.
3. Evidence and judgement failure: examples that do not prove claims, weak comparison, unsupported evaluation.
4. Precision failure: recurring paraphrase, grammar, reference or sentence-control errors that distort meaning.
5. Polish: style improvements that matter only after the higher-level decisions are stable.

Fixing lower-level polish while a student repeatedly answers the wrong question is like tuning the sound system of a vehicle with a broken steering mechanism. The work may be real, but it is not the current bottleneck.

Paper 1: turn essay feedback into future decisions

Paper 1 feedback becomes most useful when it is moved backwards from the finished essay into the decisions that produced it. The detailed examination mechanics belong in the Paper 1 owner; here the concern is transfer.

Question choice feedback

If a teacher says you chose a question for which you lacked breadth, do not merely promise to “choose better.” Reconstruct the choice. What attracted you? Familiar keywords? One strong example? A memorised essay? Then create a selection gate: before committing, generate a provisional thesis, two supporting routes, one serious challenge and enough specific evidence to sustain them. If one element is missing, the apparent familiarity may be deceptive.

Introduction feedback

If the introduction was vague, identify whether the problem was definition, scope, criteria or thesis. A stronger rewrite should not just sound better; it should make the later body easier to control. Test the repaired introduction by asking someone to predict your paragraph structure from it.

Paragraph feedback

When a paragraph is marked “descriptive,” highlight every sentence by function: claim, mechanism, evidence, implication, qualification, judgement. If most of the paragraph is evidence, the fix is not to add a concluding sentence mechanically. The fix is to make the reasoning chain visible.

Evidence feedback

For every criticised example, ask four questions: What exact claim was it meant to support? What mechanism links the example to that claim? What limitation does the example have? Would a different example prove the same point more directly? This converts “learn more examples” into evidence selection.

Evaluation feedback

Evaluation improves when the student knows what is being compared. Useful criteria include scale, durability, reversibility, distribution, feasibility, opportunity cost, time horizon and conditions. The criterion must fit the question; it should not become a generic checklist pasted onto every essay.

Conclusion feedback

A conclusion that introduces a new position often reveals that the thesis never stabilised. Compare the first and last judgement. If they differ, decide whether the essay developed legitimately or drifted. The repair may belong at the planning stage rather than in the conclusion itself.

Paper 2: turn comprehension feedback into future decisions

Paper 2 feedback is often highly local: one word was too broad, one inference missed a qualifier, one summary point duplicated another. The danger is to fix only the local answer. Instead, extract the reading or answer-construction rule that failed. Detailed mechanics belong in the Paper 2 owner.

Meaning and reference feedback

If a pronoun, phrase or relationship was misunderstood, trace the reference chain in the passage. Then test the skill on a new paragraph containing a different reference structure. The transferable rule is not “remember that this pronoun referred to X.” It is “when a reference is ambiguous, test each candidate against grammar, logic and surrounding claims.”

Paraphrase feedback

When an answer is marked too close to the passage, do not replace individual words with random synonyms. First state the underlying relationship in simpler language, then rebuild the sentence. When an answer is marked inaccurate, compare the semantic boundaries: did you make the claim stronger, weaker, broader or narrower than the passage?

Inference feedback

Write the evidence and inference on separate lines. If the inference cannot be justified from the evidence without importing outside assumptions, it is too speculative. Practise with short extracts so that the evidence-to-inference jump becomes visible.

Summary feedback

Tag each candidate point by function before writing. Eliminate duplication before compressing language. If you repeatedly miss a type of point, inspect your selection rule. If you repeatedly overwrite, practise compression separately from selection.

Application feedback

For every application paragraph, make the bridge explicit: passage claim → chosen context → concrete evidence → explanation of fit or mismatch → judgement. If the paragraph could survive after removing the passage claim, it may be general knowledge rather than application.

How to use a teacher consultation

A consultation is most productive when it begins with a precise uncertainty. Bring the marked script and a short hypothesis. Instead of asking “How do I improve my essay?”, ask “I think these three paragraphs were marked descriptive because I gave examples before explaining the mechanism. Is that the main failure, or is the claim itself too broad?”

• Ask the teacher to identify the highest-return error, not every possible improvement.
• Ask for one contrast: what would a stronger decision have looked like?
• Ask what evidence would show that the problem is fixed in a new task.
• Repeat the instruction back in your own words.
• Leave with a small drill and an unseen transfer test.

The consultation should end with an experiment, not only an explanation.

The rewrite protocol: correct less, learn more

Full rewrites are useful only when the task is to integrate many repairs. Often a partial rewrite is more diagnostic because it targets the failed decision without letting familiarity do all the work.

1. Choose one high-priority error family.
2. Rewrite only the smallest section needed to demonstrate the repair.
3. Annotate what changed and why.
4. Remove the old script.
5. Attempt a similar but unseen prompt.
6. Compare the new performance with the replacement rule, not with the beauty of the rewrite.

A polished rewrite can be satisfying. A successful unseen transfer is more valuable.

Feedback from models: use comparison, not imitation

Model essays and model answers are another form of feedback. Their value comes from contrast. Compare decisions: how was the question narrowed, where was evidence placed, how was a counterargument handled, how did the writer qualify a claim, how was a passage idea applied? Then close the model and recreate the principle on a different task.

Do not treat a model as a script to memorise. The dedicated GP model essays guide explains how to learn from exemplars without becoming dependent on them.

When feedback conflicts

Different teachers may emphasise different things because they are responding to different scripts, different stages of development or different interpretations of the same weakness. Do not resolve apparent conflict by choosing the advice you prefer.

1. Write both pieces of feedback exactly.
2. Identify the context in which each was given.
3. Ask whether they operate at different levels. “Be concise” and “explain more” can both be correct if unnecessary background is long while the causal link is short.
4. Find the common performance criterion underneath them.
5. Test both interpretations on a fresh task.
6. If the conflict remains important, take the two examples to a teacher and ask for the boundary condition: when does each instruction apply?

Three students, three feedback problems

Maren keeps receiving comments about weak development. Her first reaction is to add more examples. When she classifies the feedback, she discovers that her examples are already plentiful; the missing step is causal explanation. Her replacement rule becomes: after every example, state what changed, for whom, and why that change proves the paragraph claim. She practises on five isolated claims before returning to a full essay.

Iona loses comprehension marks through answers that are “almost right.” Her problem is not general reading ability. She broadens narrow passage meanings when paraphrasing. Her new routine is to compare the answer with the source phrase and ask whether any condition, degree or relationship has been weakened or expanded. She tests the rule on short unseen extracts before attempting a full Paper 2.

Leonie knows why marks were lost after every review, yet the same execution failures return under time. Her ledger shows that the error is not knowledge but state-dependent performance. She moves the repair into timed work: fixed checkpoints, one recovery rule when behind schedule, and a deliberate final-minute checking order. Her success measure becomes recurrence under pressure, not quality during untimed correction.

The 24-hour, 7-day and 21-day feedback cycle

Within 24 hours

Capture the feedback, classify the top two error families, write replacement rules and complete one short repair drill. Do not attempt to repair the entire subject in one sitting.

Within 7 days

Run an unseen transfer task. It can be a plan, paragraph, short extract, summary drill or timed section. Record recurrence.

Within 21 days

Revisit the error family in a larger timed task or mock. If it survives, change the diagnosis rather than merely repeating the same correction. The GP practice papers and mock exams guide explains how to use simulation for diagnosis rather than endurance alone.

A 12-week feedback-to-performance programme

• Weeks 1–2: build the ledger from recent marked work; identify the three most recurrent high-cost error families.
• Weeks 3–4: isolate the first error family with short drills and one unseen transfer test each week.
• Weeks 5–6: integrate the repair into a larger Paper 1 or Paper 2 section under moderate time pressure.
• Weeks 7–8: shift to the second error family while maintaining the first through spaced checks.
• Weeks 9–10: run full or near-full simulations; track whether old errors return under fatigue or time.
• Weeks 11–12: narrow the ledger to the remaining live constraints and connect them to final revision priorities.

This programme should not replace the broader GP revision architecture. It supplies the feedback evidence that helps revision decide what deserves attention.

60 feedback case labs

The following cases are deliberately short. Use them to practise converting a comment into a cause, a repair and an unseen test.

Case 1: “Too descriptive”

The paragraph contains facts and examples but little mechanism. Repair by writing the claim first, then forcing each example to answer “how does this make the claim true?” Test on a different topic using only one example.

Case 2: “Not answering the question”

The student discussed the broad theme but ignored a qualifier. Repair by turning every important word in the prompt into a planning checkpoint. Test with five new prompts containing different qualifiers.

Case 3: “Weak thesis”

The introduction lists both sides without deciding how they should be weighed. Repair by stating a criterion and a provisional judgement. Test whether the body paragraphs can be predicted from the thesis.

Case 4: “Example not relevant”

The example matches the topic but not the claim. Repair by writing the mechanism before selecting the example. Test with three examples from unrelated domains and reject those that cannot prove the mechanism.

Case 5: “Needs evaluation”

The paragraph ends after presenting evidence. Repair by comparing significance under one explicit criterion. Test on a new paragraph where the counterexample is stronger than the original claim.

Case 6: “Overgeneralised”

A specific example is used to support a universal claim. Repair with boundary language: where, when, for whom and under what conditions? Test by finding one case that would break the original generalisation.

Case 7: “Counterargument is tokenistic”

The opposing view appears only to be dismissed. Repair by stating the strongest plausible version of the challenge and the condition in which it matters. Test whether the final judgement changes at all after taking it seriously.

Case 8: “Paragraph drifts”

The paragraph begins with one claim and ends on another. Repair by writing a one-line paragraph contract before drafting. Test by checking whether every sentence serves that contract.

Case 9: “Too much background”

Context consumes space before analysis begins. Repair by limiting background to information necessary for the mechanism. Test by deleting each sentence and asking whether the argument becomes harder to understand.

Case 10: “Conclusion repeats introduction”

The essay ends without showing what the argument established. Repair by synthesising the criteria that decided the judgement. Test by writing the conclusion from the body plan rather than copying the introduction.

Case 11: “Vague language”

Words such as “things,” “a lot,” “bad” or “helps” hide the mechanism. Repair by replacing each with the exact actor, change and consequence. Test under a five-minute editing limit.

Case 12: “Sentence too long”

Several logical relationships are packed into one sentence. Repair by separating the main claim from qualification and evidence. Test whether each sentence can be read once without losing its grammatical centre.

Case 13: “Repetition”

The same idea is restated with different words. Repair by labelling sentence functions and removing duplicates. Test by compressing the paragraph by twenty per cent without losing reasoning.

Case 14: “Example dump”

Several examples appear in succession without explanation. Repair by keeping one or two and increasing mechanism and comparison. Test whether removing one example weakens the argument; if not, it was probably decorative.

Case 15: “Knowledge inaccurate”

A factual claim is uncertain or overstated. Repair by recording source, date, scope and limit in the evidence bank. Test by explaining the example without notes and stating one limitation.

Case 16: “Too one-sided”

The student never pressure-tests the preferred position. Repair by generating the strongest countercondition before drafting. Test on a topic where the opposing side is initially more familiar.

Case 17: “No clear criterion”

The essay uses words like “important” without defining importance. Repair by specifying what counts: scale, durability, fairness, feasibility or another question-relevant standard. Test whether two examples can be ranked using it.

Case 18: “Paragraph conclusion unsupported”

The last sentence claims more than the paragraph proved. Repair by matching the strength of judgement to the evidence. Test by underlining every absolute word and asking whether it is warranted.

Case 19: “Question choice too ambitious”

The student understands the issue but lacks enough specific evidence. Repair with a pre-commitment gate: thesis, two routes, one challenge, evidence for each. Test during timed question selection.

Case 20: “Plan too long”

The plan becomes a miniature essay. Repair by planning decisions rather than sentences: thesis, paragraph claim, example, evaluation move. Test with a fixed planning ceiling and compare completion.

Case 21: “Misread reference”

A pronoun or phrase is attached to the wrong antecedent. Repair by testing candidate references against grammar and logic. Test on a fresh passage with nested references.

Case 22: “Copied language”

The answer repeats passage wording without demonstrating understanding. Repair by stating the relationship in plain language before writing the final response. Test on five short phrases with no thesaurus-style substitution.

Case 23: “Paraphrase changes meaning”

The replacement word is broader or stronger than the original. Repair by comparing semantic boundaries rather than vocabulary difficulty. Test by classifying paraphrases as equal, broader, narrower or different.

Case 24: “Inference too speculative”

The answer imports assumptions not supported by the text. Repair by writing evidence and inference separately. Test whether each inference can be defended using only the passage.

Case 25: “Answer incomplete”

The response captures one part of a two-part relationship. Repair by translating the question into required answer components before searching the passage. Test on questions with contrast, cause-and-effect and comparison structures.

Case 26: “Answer too broad”

The student gives a correct general idea but misses the passage’s specific distinction. Repair by locating the smallest wording that narrows the meaning. Test with answers that differ only in scope.

Case 27: “Wrong tone”

The student labels attitude from one emotional word. Repair by combining diction, contrast and target of the attitude. Test with passages where tone shifts mid-paragraph.

Case 28: “Summary misses key points”

The student compresses too early and overlooks categories. Repair by selecting and grouping before paraphrasing. Test with short passages where several examples support one underlying point.

Case 29: “Summary duplicates points”

Two sentences express the same underlying idea. Repair by labelling each candidate point before drafting. Test by merging overlapping points without losing distinct consequences.

Case 30: “Summary too wordy”

The student preserves examples and explanation that are not required. Repair by keeping the proposition and deleting illustration unless the illustration itself is the point. Test with progressive compression.

Case 31: “Application is generic”

The response discusses society without linking back to the passage claim. Repair by naming the claim at the start of each application unit. Test whether deleting the claim makes the paragraph collapse; if not, the bridge is too weak.

Case 32: “Application example is anecdotal”

A personal observation is treated as proof of a broad social claim. Repair by identifying what the anecdote can and cannot establish. Test by pairing it with broader evidence or narrowing the judgement.

Case 33: “Application lacks balance”

The student applies only passage ideas that support a preferred view. Repair by selecting at least one idea that creates tension. Test whether the final judgement survives the strongest counterexample.

Case 34: “Application evidence is dated”

The example may no longer represent the current context. Repair by verifying date and present relevance. Test whether the mechanism still operates rather than assuming recency from familiarity.

Case 35: “Application repeats passage”

The student paraphrases the author but adds little contextual analysis. Repair by requiring a concrete local or comparative mechanism after each passage claim. Test on a context the student has not rehearsed.

Case 36: “Slow start”

Too much time is spent waiting for confidence. Repair with a start protocol: decode, decide, outline, begin. Test under several short timed starts rather than one full paper.

Case 37: “Over-editing early answers”

The student spends disproportionate time polishing before later questions are attempted. Repair with a first-pass threshold and return list. Test using a visible checkpoint.

Case 38: “No time to check”

The issue may be earlier allocation, not checking skill. Repair by tracing where time was lost and assigning a minimum checking reserve. Test the reserve under a timed section.

Case 39: “Panic after a difficult question”

One uncertain item disrupts the rest of the paper. Repair with a recovery script: mark uncertainty, make the best bounded decision, move, return later. Test with deliberately inserted difficult items.

Case 40: “Performance collapses late”

Fatigue causes precision errors. Repair by moving high-risk checks to planned moments and building stamina gradually. Test whether the same error family rises with task duration.

Case 41: “Teacher says ‘be more nuanced’”

Translate the abstract instruction into conditions. Ask: where would my claim fail, for whom, and under what timeframe? Test by rewriting three absolute claims as conditional judgements.

Case 42: “Teacher says ‘go deeper’”

Depth may mean mechanism, implication or comparison. Identify which layer is missing. Test by adding only one missing layer and seeing whether the paragraph becomes more explanatory rather than merely longer.

Case 43: “Teacher says ‘be concise’”

Find low-function sentences: repeated setup, obvious transitions, duplicated evidence. Remove them while protecting reasoning. Test by cutting fifteen per cent of a paragraph without losing any logical step.

Case 44: “Teacher says ‘explain more’”

Do not simply lengthen the paragraph. Identify the invisible link between evidence and claim. Add the causal or comparative step and test whether a reader can now follow the inference without guessing.

Case 45: “Teacher says ‘use more evidence’”

Check whether the problem is quantity or specificity. One vague example and three vague examples remain weak. Build an evidence unit containing who or what, when, what happened, mechanism and limitation.

Case 46: “Teacher says ‘organise better’”

Find whether the disorder is at essay, paragraph or sentence level. Repair the smallest level that causes the rest. Test by outlining the logic after writing; if the outline is incoherent, the structure is still unstable.

Case 47: “Teacher circles grammar repeatedly”

Group the circles by error family instead of correcting each sentence independently. If most are agreement, tense, reference or boundary errors, practise that family in short editing drills and then retest in timed writing.

Case 48: “Teacher writes ‘awkward’”

Determine whether the issue is grammar, collocation, register or overloaded syntax. Rewrite for clear meaning first, then compare. Test the same sentence pattern with new content rather than memorising the corrected sentence.

Case 49: “Teacher says ‘too sweeping’”

Identify the population, time, place or condition that the claim wrongly universalises. Repair by narrowing or qualifying. Test by asking what evidence would falsify the revised claim.

Case 50: “Teacher says ‘link back to question’”

The paragraph may be relevant in topic but not in judgement. Repair by restating the paragraph’s significance using the exact criterion of the question. Test whether the link changes if the qualifier changes.

Case 51: Strong correction, weak new essay

The student can follow feedback when the original task is visible but does not retrieve it independently. Repair with closed-book transfer: write the replacement rule from memory, then use it on a new prompt.

Case 52: Improvement in class, relapse in exam

The skill is state-dependent. Move practice into timed, uncertain conditions gradually. Track whether the error appears only after fatigue, difficult question choice or time pressure.

Case 53: Too many corrections at once

The student tries to remember ten rules during drafting. Repair by selecting one or two high-cost constraints for the next task. Keep the rest in the ledger, not in working memory.

Case 54: Feedback feels obvious

The student says, “I knew that.” Knowledge after the event is not the same as retrieval during the event. Test by predicting the error before the next task and setting a checkpoint where the rule must be used.

Case 55: Feedback feels too abstract

A comment such as “more sophisticated” is not actionable. Ask for one observable contrast between the current response and a stronger one. Convert the contrast into a decision rule and a small drill.

Case 56: Different teachers give different advice

Map the context and level of each comment. One may target concision while another targets missing reasoning. Test the boundary condition rather than treating the advice as mutually exclusive.

Case 57: Mark rises once, then falls

One improved score may reflect topic fit or task variation. Track the repaired error directly across several unseen tasks. Stability matters more than a single peak.

Case 58: Student copies model language

The output looks polished but transfer is fragile. Strip away the model wording, extract the underlying move, and apply it to a different issue. Test whether the student can explain why the move works.

Case 59: Student avoids difficult feedback

Easy corrections create visible progress while the binding constraint survives. Use the priority ladder and schedule the highest-cost recurring error first, even if it is uncomfortable.

Case 60: Student cannot tell what improved

The review has no baseline. Compare recurrence by error family, not only total marks. A disappearing task-interpretation error is meaningful even if a new evidence weakness becomes visible next.

40 feedback-transfer drills

1. Take five teacher comments and rewrite each as an observable if–then–check rule.
2. Sort twenty past comments into no more than eight error families.
3. Choose one “too vague” paragraph and underline every noun whose reference is unclear.
4. Take one irrelevant example and explain why topic similarity is not enough to make it evidence.
5. Write three different thesis statements for the same prompt, each using a different criterion.
6. Turn five absolute claims into conditional claims without making them empty or evasive.
7. Take one descriptive paragraph and label claim, mechanism, evidence, implication and judgement.
8. Remove one-third of the examples from a paragraph and strengthen the explanation of what remains.
9. Take a teacher comment that says “deeper” and identify whether depth means mechanism, implication, comparison or condition.
10. Write a paragraph where the counterargument genuinely changes the final judgement.
11. Choose ten prompts and circle the word most likely to be ignored under pressure.
12. Plan three essay questions in short form and reject one using a question-choice gate.
13. Compare the first and final judgement of an old essay and identify drift.
14. Rewrite only the weakest paragraph of a marked essay, then attempt a new paragraph on a different topic using the same repair.
15. Build an evidence card with source, date, mechanism, use and limitation for one example.
16. Take five passage phrases and paraphrase the relationship, not individual words.
17. Classify ten paraphrases as equivalent, broader, narrower or changed.
18. For five inference questions, write evidence and inference on separate lines.
19. Trace every pronoun in one dense paragraph to its likely antecedent.
20. Take a summary passage and select points before writing any paraphrase.
21. Mark possible duplicate summary points before compressing them.
22. Compress one 40-word explanation to 25 words, then 18, preserving the central relationship.
23. For one application paragraph, draw arrows from passage claim to context evidence to judgement.
24. Write two application paragraphs using the same passage claim in different contexts.
25. Find one case where the passage claim does not apply cleanly and explain the boundary.
26. Run a ten-minute timed start routine on three different days.
27. Insert a deliberately difficult question into practice and rehearse the recovery script.
28. Mark time checkpoints on a practice paper and record actual arrival times.
29. Design a final-minute checking order based on your highest-cost recurrent errors.
30. Correct one grammar family across ten sentences, then write a timed paragraph without monitoring every rule consciously.
31. Compare two teacher comments that seem contradictory and identify the different levels they target.
32. Before a consultation, write your own diagnosis in one sentence and ask the teacher to challenge it.
33. After a consultation, explain the repair to someone else without looking at notes.
34. Close a model essay and reconstruct only its argument architecture from memory.
35. Use the same architecture on a question from another domain.
36. Choose one error scored 3 on your recurrence scale and run three short transfer tests in a week.
37. Choose one error scored 1 and test whether it disappears under time pressure.
38. At the end of a mock, predict the three feedback comments you are most likely to receive before seeing the script.
39. Compare your prediction with the actual marking and update your self-monitoring model.
40. At the end of three weeks, remove solved errors from the active list and promote the next live constraint.

A 100-question self-audit for marked scripts

You do not need to answer all one hundred after every paper. Use the relevant cluster when a script is returned.

Task and scope

1. Did I answer the exact task rather than the broad topic?
2. Which word in the question controlled scope most strongly?
3. Did I ignore a qualifier?
4. Did I answer only part of a comparison?
5. Did I assume a definition that needed examination?
6. Was my chosen interpretation defensible?
7. Did the thesis match that interpretation?
8. Did later paragraphs drift from it?
9. Did the conclusion change the task?
10. What decision rule would prevent the same scope failure?

Argument and reasoning

11. Was my thesis genuinely arguable?
12. Did I state criteria for judgement?
13. Did each paragraph have one controlling claim?
14. Was the mechanism between claim and evidence explicit?
15. Did I mistake description for analysis?
16. Did I repeat instead of develop?
17. Did a counterargument change the reasoning?
18. Were conditions and limits explicit?
19. Did the paragraph ending match what was proved?
20. Which reasoning move needs a transfer drill?

Evidence

21. Was the example accurate enough?
22. Was it specific enough?
23. Was it current enough for the claim?
24. Did I explain the mechanism?
25. Did I rely on name-dropping?
26. Did I overgeneralise from one case?
27. Would another example prove the point more directly?
28. Did I distinguish evidence from illustration?
29. Did I state a relevant limitation?
30. What evidence gap should enter my study plan?

Evaluation

31. What exactly was I evaluating?
32. Which criterion did I use?
33. Was the criterion appropriate to the question?
34. Did I compare scale?
35. Did I compare time horizons?
36. Did I consider affected groups?
37. Did I identify boundary conditions?
38. Did I make a judgement before enough evidence?
39. Did I postpone all judgement until the conclusion?
40. What evaluation move should I rehearse next?

Language

41. Which language errors actually changed meaning?
42. Which error family repeated most?
43. Were sentence boundaries stable?
44. Were references clear?
45. Did vocabulary become vague under pressure?
46. Did complex syntax improve or obscure meaning?
47. Did I use inappropriate register?
48. Did I repeat key words because I lacked alternatives or because precision required them?
49. Which language repair can be isolated in a short drill?
50. Can I execute it while thinking about content?

Paper 2 reading and answers

51. Did I locate the correct passage evidence?
52. Did I understand the relationship between ideas?
53. Did I track references accurately?
54. Did my paraphrase preserve degree?
55. Did it preserve cause and effect?
56. Did it preserve comparison?
57. Did I import outside assumptions?
58. Did I answer every required component?
59. Was my answer broader than the passage?
60. What reading rule should transfer to the next passage?

Summary

61. Did I select before paraphrasing?
62. Did I miss an entire category of point?
63. Did I duplicate ideas?
64. Did I preserve examples that should have been compressed?
65. Did I remove necessary relationships?
66. Was the wording concise but accurate?
67. Did I manage the word limit strategically?
68. Which stage caused the main loss: selection, grouping, paraphrase or compression?
69. Can I isolate that stage in practice?
70. Did the problem recur across earlier scripts?

Application

71. Did I state the passage idea accurately?
72. Did I choose a context that genuinely tested it?
73. Was my evidence concrete?
74. Did I explain the bridge between text and context?
75. Did I include a mismatch or boundary?
76. Was my judgement proportionate?
77. Did I rely on anecdote?
78. Did I repeat the passage instead of applying it?
79. Could the paragraph stand without the passage idea?
80. What transfer test would show improvement?

Execution

81. Where did I lose the most time?
82. Was the loss caused by uncertainty, perfectionism or weak process?
83. Did I reach planned checkpoints?
84. Did one difficult item disrupt later work?
85. Did I leave enough time to check?
86. Was my checking order efficient?
87. Did fatigue change my error pattern?
88. Did I recover after noticing a mistake?
89. Which execution rule should be tested under time?
90. Did I distinguish knowledge failure from performance failure?

Transfer

91. What are the two highest-cost comments on this script?
92. What is the likely cause of each?
93. What replacement decision rule will I use?
94. What short rehearsal will isolate it?
95. What unseen task will test transfer?
96. When will I run that test?
97. How will I score recurrence?
98. What evidence would prove my diagnosis was wrong?
99. When will I remove this error from the active ledger?
100. What becomes the next constraint when it is solved?

What not to do with feedback

• Do not copy every correction into a notebook without classifying it.
• Do not treat marks as the only feedback signal.
• Do not respond to every weak example by collecting more examples.
• Do not rewrite whole essays repeatedly when one decision is the real failure.
• Do not memorise teacher wording as if it were a universal template.
• Do not confuse understanding a comment with being able to act on it under time.
• Do not keep solved errors permanently in the active list.
• Do not chase style improvements while task interpretation is unstable.
• Do not turn the ledger into a punishment record. It is a control system.
• Do not wait for a teacher to notice every recurrence. Learn to predict your own likely feedback.

How feedback connects to the rest of the JC2 system

Feedback is one input into a larger final-year control system. Use the How to Study for GP guide to build the weekly learning loop. Use the GP Revision Guide to decide what should be retrieved and repaired. Use How to Improve GP Grades when the main task is identifying the current binding constraint. Use the Prelim to A-Level Strategy when marked preliminary examinations must be converted into final-runway decisions.

The purpose of this page is narrower and therefore more demanding: it asks whether feedback actually changes the next unseen performance.

The independence test

Feedback has done its job when the student no longer needs the original comment to trigger the improved decision. The rule has become part of performance.

• You can predict likely feedback before the script is returned.
• You can identify the probable cause of an error without waiting for a full explanation.
• You can choose one high-return repair rather than correcting everything equally.
• You can apply the repair to an unfamiliar question or passage.
• You can detect recurrence under time pressure.
• You can retire solved errors and move attention to the next live constraint.

That is the final purpose of marked work. The paper should gradually become less necessary because the student has internalised the diagnostic process.

Final principle: every comment must earn its way into the next task

A returned script is valuable only if something in the next performance changes. Capture the evidence. Classify the error. Find the cause. Write a replacement rule. Rehearse it. Transfer it. Track recurrence. Then move on.

The strongest JC2 feedback system is not the one with the most annotations. It is the one in which fewer old errors survive each new unseen task.

Return to the JC2 General Paper hub for the complete final-year route.