IGCSE composition practice becomes useful when it changes what a student can do on an unseen task without help. Repeating full compositions can produce familiarity, stamina and confidence, but it can also rehearse the same weaknesses. A stronger system moves deliberately from understanding to control, from control to variation, from variation to independence, and from independence to timed examination transfer.
This guide belongs to SETC’s Cambridge IGCSE First Language English Composition adapter. It stays inside narrative and descriptive composition. It does not own Directed Writing, source transformation or functional-writing practice. It also does not replace SETC’s generic practice-engineering, feedback, error-analysis or creative-writing owners. This page owns the composition-specific training architecture: what to practise, in what order, under what conditions, and how to know when a skill is ready for unseen examination use.
Cambridge’s 2027–2029 First Language English syllabus continues to assess composition within Paper 2, with revised wording and marking criteria for the new cycle. Candidates and teachers should consult current official syllabus and specimen materials for exact requirements. The practice system here is built around durable performance qualities: relevance, structure, development, style, accuracy, independence, timing and transfer.
Practice is not one condition
A student can write a strong paragraph after a teacher explains exactly what to do. That shows supported performance. The same student may struggle when the prompt changes. Another may succeed on a familiar prompt but fail when time is added. These are different stages of learning.
A useful practice system distinguishes learning, guided practice, independent practice, transfer and simulation. Confusing them creates false confidence. A model-assisted paragraph is not yet examination-ready simply because it is excellent.
The IGCSE Composition Practice Ladder
- Understand the mechanism. Learn what good narrative causality, descriptive movement, viewpoint, paragraph function or language control actually does.
- Perform with support. Use prompts, models, checklists or teacher questions.
- Perform with reduced support. Remove one scaffold at a time.
- Vary the surface. Change topic, setting, viewpoint or task while keeping the underlying method.
- Mix the demand. Require the learner to identify which method is needed.
- Transfer independently. Use an unseen prompt without model support.
- Add realistic time. Protect quality while decision cost rises.
- Simulate. Complete the whole examination task under authentic conditions.
- Diagnose. Identify the first weak link from the simulation.
- Return to repair. Do not stay permanently at full-paper level.
Maren, Iona and Leonie inside the practice system
Maren monitors architecture. She asks whether practice improves control over narrative change, descriptive movement and paragraph function rather than merely producing nicer sentences.
Iona monitors evidence of transfer. She asks whether the learner can explain why a choice works and reproduce the principle when the prompt changes.
Leonie monitors execution. She asks when to introduce time, how much support to remove and whether performance remains stable under uncertainty.
Stage 1: mechanism learning
Do not begin every skill with a full composition. If the target is narrative turning points, study and build turning points. If the target is descriptive focal movement, practise moving from wide to close focus. If the target is ending control, write endings. Full compositions are expensive; small tasks reveal the mechanism more clearly.
Mechanism learning should make invisible decisions visible
Ask why one paragraph works better than another. Label narrative cause and consequence. Mark descriptive movement. Compare an ending that resolves pressure with one that merely stops. Examine how one sentence establishes viewpoint. The student should be able to name the decision before trying to automate it.
Stage 2: supported production
Support can include paragraph frames, scene maps, question prompts, model comparison, vocabulary banks or teacher cues. The purpose is to allow successful performance while the mechanism is still fragile.
Support should be designed to disappear. A scaffold that remains forever becomes part of the task instead of a bridge to independence.
Stage 3: scaffold fading
Remove support in small steps. Full paragraph plan becomes paragraph-job labels. Detailed checklist becomes three cues. Teacher questions become a self-check card. Model text becomes only a title or structural outline. The learner should feel slightly more decision load without losing the method entirely.
Do not confuse successful scaffolding with independent mastery
If the student succeeds only when the teacher asks “What changes here?” or “Where is the reader standing?”, the method still partly lives in the teacher. Count prompts. Independence becomes measurable when prompts reduce without quality collapse.
Stage 4: controlled variation
Keep the deeper skill constant and change one surface feature. Narrative causality can be practised across school, family, travel and competition settings. Descriptive movement can be practised across market, station, coastline and empty building. The learner discovers that the method is not tied to one prompt.
Change one variable first
If topic, viewpoint, genre demand, time and language difficulty all change simultaneously, failure becomes hard to diagnose. Early transfer should vary one or two conditions. Later practice can combine them.
Stage 5: discrimination
Blocked practice tells the student which method to use. Examination prompts do not. Mix narrative openings, descriptive routes, dialogue decisions, sensory selection and ending problems. Ask the learner first to identify what kind of control is required.
Recognition is part of performance
A student who knows ten writing techniques but cannot recognise when to use each one is not examination-ready. Practice should therefore include method selection, not only method execution.
Stage 6: independent transfer
Give an unseen prompt without models, hints or vocabulary lists. Allow normal planning. Observe where the student hesitates. The goal is not a perfect script. The goal is to see whether the operating system transfers.
The independent first attempt should come before model comparison
Showing a model first can make later performance look stronger than it is because the student borrows its structure and phrasing. Attempt first, compare second. Models then become diagnostic tools.
Stage 7: timed segments
Do not jump from untimed work directly to full timed composition. Time the parts that matter: prompt choice, planning, opening, one body paragraph, climax, descriptive route, ending and checking. Identify where quality first falls.
Time should expose the bottleneck
If planning takes too long, repair planning. If vocabulary search slows drafting, practise lexical retrieval. If the ending collapses, protect time and pre-plan changed state. A global stopwatch cannot tell you which decision is expensive.
Stage 8: partial simulation
Combine realistic prompt selection, planning and a substantial portion of the task without yet demanding the full examination. Partial simulations help bridge segment timing and full-paper pressure.
Stage 9: full simulation
A full simulation is a measurement event as much as a practice event. It should answer a question: Does quality survive the clock? Does the learner choose well? Does the narrative finish? Does description maintain movement? Does language accuracy fall late?
Stage 10: mock-to-repair
After simulation, classify the main losses: prompt choice, architecture, development, viewpoint, language precision, grammar under load, timing, editing or recovery. Then isolate the dominant mechanism before the next full simulation.
The repair–practice–transfer–simulate loop
Repair: teach or rebuild the weak mechanism. Practice: repeat under controlled conditions. Transfer: change surface features and reduce support. Simulate: test under examination conditions. If simulation fails, return to the earliest unstable stage.
Narrative practice should not be only full stories
Write five openings for one prompt. Build three turning points from the same setup. Rewrite a weak climax so it grows causally. Write an ending from a supplied middle. Compress a 600-word plot into a five-event causal chain. Change viewpoint while preserving events. These tasks teach decisions more efficiently than another full story every lesson.
Descriptive practice should not be only full scenes
Describe the same image from three positions. Write only wide-to-close movement. Remove thirty per cent of details without losing atmosphere. Rewrite a static paragraph so attention moves. Describe the same place before and after rain. Write one paragraph dominated by sound rather than sight.
Practice prompt choice separately
Give a set of narrative and descriptive options. Allow five minutes to rank them. Require reasons based on structure, viewpoint, detail supply, ending control and time risk. No full composition is needed. Selection improves through repeated comparison.
Practice planning separately
Planning-only sessions can cover many prompts quickly. For narrative: opening state, pressure, causal movement, turning point, changed state. For description: viewpoint, route, dominant sensory channels, paragraph jobs, final view. Compare plans for finishability.
Practice paragraph roles
Give a composition and ask the student to label each paragraph’s job. Then remove or merge any paragraph that duplicates function. Later, plan new pieces using role labels before content.
Practice openings and endings as a pair
Write an opening state and ending state before drafting the middle. This protects causal movement and prevents a rushed ending that does not belong to the beginning.
Practice causal chains
Use five boxes: state → disturbance → choice → consequence → changed state. Then vary the setting while preserving the logic. This teaches narrative architecture without memorising a story.
Practice descriptive routes
Use route sketches: outside → threshold → interior → close detail → final wide view; or foreground → middle distance → background → return to foreground. Transfer the same route logic to different settings until movement becomes controllable.
Practice sensory hierarchy
Write scenes using only the two most relevant senses, then add a third if it genuinely contributes. This breaks the five-senses checklist habit and improves selection.
Practice vocabulary through function
Build lexical sets for movement, texture, sound, uncertainty, tension, stillness, decay, brightness, crowding and distance. Include collocations and sentence contexts. Vocabulary becomes retrievable because it is organised by writing function.
Practice sentence structures through meaning
Use short sentences for isolation or decisive action, subordination for condition and concession, parallel structure for pattern, and coordinated clauses for simultaneous movement. Sentence practice should connect structure to effect.
Practice editing on delayed drafts
Students often miss errors immediately after writing because they still remember intended meaning. Revisit a draft later. First edit structure and development, then sentence meaning, then language accuracy. Delay improves detection.
Practice self-detection before teacher correction
Before feedback, ask the learner to mark three areas they distrust. The tutor then compares self-detection with actual problems. Over time, the student should become better at predicting where control failed.
Practice feedback as a future cue
A correction is incomplete until it changes the next task. Convert feedback into a trigger-action pair: “If my opening reaches 150 words before pressure appears, compress.” “If I describe three close details in a row, re-establish spatial orientation.” “If two characters share the same pronoun, check reference.”
Keep only three active risks
A composition can contain many imperfections. The learner cannot monitor all of them at once. Select three active risks based on frequency and impact. Once one stabilises across several unseen tasks, replace it.
Use spaced transfer checks
Retest a repaired skill after one week and again after several weeks. Immediate improvement may reflect fresh feedback rather than durable learning. Spacing reveals retention.
Use interleaving once methods are understood
Mix narrative and descriptive planning, opening, paragraph and editing tasks. The learner must first identify the demand. Interleaving is useful after basic methods exist; too early, it can create confusion.
Use retrieval before review
Before showing the checklist for narrative or description, ask the student to reconstruct it. What makes an ending complete? What are the main descriptive movement routes? What are the active risks? Retrieval tells you what is actually available under examination conditions.
Use contrast sets
Compare two openings, two endings, two descriptive paragraphs or two plans. Ask which is stronger and why. Contrast builds criteria. Students learn more from explaining the difference than from copying one “good” model.
Use deliberate weak models
Not every model should be excellent. A deliberately flawed paragraph can reveal overdescription, viewpoint leak, generic atmosphere, causal weakness or language inflation. Diagnosis practice strengthens editing judgement.
Use model texts after independent thinking
Models are most useful when the learner can compare them with their own choices. Ask what the model does differently at the level of architecture, development, viewpoint and language. Avoid phrase harvesting.
Measure progress beyond marks
- Prompt choice becomes faster and more defensible.
- Plans become shorter but more useful.
- Openings establish direction sooner.
- Narrative events become more causal.
- Descriptive movement becomes more coherent.
- Viewpoint errors decrease.
- Language becomes more precise without becoming inflated.
- Endings are completed more reliably.
- Active errors transfer less often.
- Teacher prompts decrease.
- Timed quality approaches untimed quality.
- Unseen prompts feel less destabilising.
The composition practice dashboard
Track four categories: control (can the skill be performed?), transfer (does it survive new prompts?), timing (does it survive the clock?) and independence (does it survive removal of support?). A skill is examination-ready only when all four are reasonably stable.
Practice errors should be classified by first cause
One rushed ending may be caused by slow prompt choice, excessive planning, overwritten opening or slow sentence production. Diagnose the earliest repairable cause. Otherwise the student may practise endings when the real problem is time allocation earlier in the task.
Narrative practice laboratory | Thirty worked routines
1. Five-opening drill: write five different openings for one prompt and rank by direction.
2. Five-ending drill: write changed-state, quiet-aftermath, decision, circular and open endings for one story.
3. Causal-chain drill: reduce plot to five because-therefore links.
4. Stakes drill: keep event fixed, vary what character can lose.
5. Turning-point drill: write three decisions that could redirect same setup.
6. Climax-preparation drill: add two earlier details that make final event credible.
7. Viewpoint drill: rewrite scene in first-person and third-person limited, compare information access.
8. Dialogue-function drill: remove every line that does not change scene.
9. Description-compression drill: cut scenic paragraph in half without losing stakes.
10. Pace drill: mark compress/normal/expand across story plan.
11. Tense-anchor drill: identify narrative now and legitimate shifts.
12. Pronoun-risk drill: edit multi-character scene for reference.
13. Emotion-without-label drill: remove emotion adjectives, show through behaviour.
14. Plot-substitution drill: keep causal architecture, change setting completely.
15. Prompt-transfer drill: adapt same underlying conflict to three unseen prompts without forcing identical story.
16. Paragraph-role drill: label setup, complication, escalation, turning, consequence.
17. Opening-economy drill: cut 180-word opening to 90 while preserving direction.
18. Ending-reserve drill: write full middle with fixed time left for ending and assess quality.
19. Weak-example drill: identify why a model story feels dramatic but incoherent.
20. No-model drill: unseen prompt with no support.
21. Prompt-choice drill: rank narrative options by finishability.
22. Three-plan drill: generate three possible story architectures before selecting one.
23. Sentence-pressure drill: write climax with clear syntax under time.
24. Revision-hierarchy drill: fix structure before language.
25. Delayed-edit drill: revisit script after two days.
26. Feedback-cue drill: convert tutor comment into next-task trigger.
27. Timed-paragraph drill: test one body paragraph under realistic pace.
28. Partial-simulation drill: prompt choice + plan + opening + one developed middle scene.
29. Full-simulation drill: complete unseen narrative, then classify first failure.
30. Retest drill: after repair, use new prompt and no reminders.
Descriptive practice laboratory | Thirty worked routines
1. Viewpoint rotation: same place from four positions.
2. Distance drill: wide, medium, close.
3. Route drill: outside-to-inside.
4. Route drill: near-to-far.
5. Route drill: centre-to-edge.
6. Sound-dominant scene: reduce visual dependence.
7. Light-dominant scene: organise by illumination and shadow.
8. Texture-dominant scene: create physical reality through contact.
9. Sensory-pruning drill: remove forced senses.
10. Dominant-detail drill: one memorable feature per paragraph.
11. Detail-hierarchy drill: sort dominant/supporting/micro.
12. Atmosphere-without-label drill: remove mood adjectives and recreate effect.
13. Metaphor-economy drill: keep only one useful comparison per paragraph.
14. Movement-without-plot drill: observer walks through scene but acquires no story goal.
15. Temporal-change drill: same scene before/after rain.
16. Crowd-density drill: empty → busy → empty.
17. Paragraph-function drill: label each focus shift.
18. Spatial-coherence drill: draw map from finished description and see if it matches.
19. Opening-orientation drill: establish viewpoint and dominant pattern in under 100 words.
20. Ending-concentration drill: write three final images that complete route.
21. Cliché-removal drill: replace stock weather and sensory phrases with observed relations.
22. Verb-strength drill: reduce adjective load through precise motion verbs.
23. Collocation drill: test descriptive vocabulary in natural phrases.
24. Sentence-rhythm drill: vary pace according to attention movement.
25. Unseen-image drill: three-minute route plan from new image.
26. No-image drill: build scene from verbal prompt using route and hierarchy.
27. Timed-section drill: two paragraphs under realistic pace.
28. Partial-simulation drill: plan + opening + two movement paragraphs.
29. Full-simulation drill: unseen description under exam conditions.
30. Transfer-retest drill: new setting, same movement principle, no prompts.
The weekly composition training rhythm
A useful week can include one mechanism lesson, one small production task, one transfer task, one editing or feedback session and one longer independent piece. Not every week needs a full composition. The exact rhythm should follow the active risk.
The monthly composition cycle
- Week 1: diagnose and repair one mechanism.
- Week 2: practise and vary.
- Week 3: transfer independently and add moderate time.
- Week 4: simulate, review and update active risks.
When to increase difficulty
Increase difficulty when the method is stable under current conditions. Difficulty can mean more unfamiliar prompt, less support, tighter time, more complex viewpoint or higher language demand. Change one dimension deliberately.
When to reduce difficulty
If the learner cannot identify the mechanism, is making random decisions or repeats the same failure under full simulation, step back. Reduce task size or support one stage. This is not lowering expectations; it is isolating the repair.
When to stop doing full compositions
Temporarily reduce full compositions when the same error cluster repeats, feedback is not transferring, timing failure hides deeper mechanism or the learner is becoming better at surviving weaknesses rather than repairing them.
When to return to full compositions
Return after the weak mechanism succeeds in smaller tasks, changed contexts and independent transfer. The full composition then tests integration.
Mock composition analysis
After a full timed task, review in this order: prompt choice, prompt fit, architecture, development, viewpoint, language precision, accuracy, completion, checking. Identify the earliest weakness that caused later losses. Then choose the next practice block.
Score stability matters more than one excellent script
Track several unseen compositions. Compare the best, typical and weak performance. The aim is to raise the floor. A candidate whose best piece is brilliant but whose weak-day piece is incoherent still has examination risk.
The performance-floor test
Use a less preferred prompt, moderate time pressure and no external support. Can the student still produce a relevant, structured, complete piece with readable language? This is a stronger readiness indicator than a polished homework composition.
The final transfer test
Give a genuinely unseen prompt from a different surface domain. No model, no vocabulary bank, no tutor cue. The learner chooses, plans, writes and checks independently. Afterwards, ask them to explain which methods they used and why. Successful transfer includes both performance and awareness.
What practice should feel like near the examination
Late-stage practice should become more representative but not become continuous simulation. Maintain short repair drills for active risks, regular unseen planning, occasional full timed compositions, retrieval of writing routines and deliberate rest. Avoid introducing large new systems unless the current one is demonstrably failing.
Final-week practice
Use light retrieval: prompt-choice criteria, narrative causal map, descriptive route options, active language risks, ending protection and checking routine. Complete one or two short representative tasks if useful. Do not exhaust the learner with repeated full compositions or late model memorisation.
Tutor audit
- Does every practice task have a specific purpose?
- Am I distinguishing learning from testing?
- Are full compositions separated by repair?
- Are scaffolds being faded deliberately?
- Is variation controlled enough to diagnose?
- Does the learner recognise which method is needed?
- Are unseen prompts used before declaring mastery?
- Is timing introduced by stage?
- Are active risks limited and tracked?
- Does feedback transfer after delay?
- Are narrative and descriptive practice balanced according to need?
- Is the performance floor rising?
Student audit
- I know what today’s practice is trying to improve.
- I can explain the method before using it.
- I can use the method when the prompt changes.
- I do not need a model before every attempt.
- I know my three active composition risks.
- I can choose a prompt for finishability.
- I can plan quickly enough.
- I can protect my ending.
- I can edit structure before vocabulary.
- I can explain what failed first after a timed task.
- I can use feedback on a new prompt without reminders.
The IGCSE Composition Practice System standard
World-class practice is not the maximum number of compositions. It is the minimum set of well-designed experiences required to make the right writing decisions independent, flexible and reliable. The student learns a mechanism, practises it with support, transfers it under variation, performs it under time and uses simulation to decide what to repair next.
Properly taught kids shine a bright light into the future. In composition practice, that light is transfer: the student can meet a prompt they have never seen, recognise what the task demands, build a controllable piece and finish it without needing the exact worksheet, model or teacher cue that taught the skill.
Where to go next
- Return to English Writing and Composition.
- Use IGCSE Narrative Openings and Endings or IGCSE Descriptive Perspective and Movement for specific craft repairs.
- Use IGCSE Composition Mistakes when progress has plateaued and the first weak link is unclear.
- Use The English Tutor Handbook | Practice Engineering for the broader practice-design mechanism.
- Use English Exam Error Analysis for cross-examination diagnosis.