English Writing and Composition · Creative Writing Handbook · How Feedback Changes Performance
SETC-ETL-0022
A model text should show the learner what good decisions look like. It should not become a script the learner carries unchanged into every new task.
Exemplars are powerful because they make invisible craft visible. Students can inspect how a writer opens, organises, develops, qualifies, transitions, integrates evidence, controls tone and finishes. But model texts can also create brittle learning when students memorise phrases without understanding the underlying move.
This tutorial develops a progression from noticing models to deconstructing them, imitating one move, varying it and transferring it into original work. It sits beneath SETC’s writing, creative-writing and exam-performance owners. The objective is not to sound like the model. The objective is to learn what the model is doing and reproduce that decision in new material.
Models make quality concrete
“Write a good introduction” is vague. A model can show a reader what direction, relevance and proportion look like on the page.
The Institute of Education Sciences’ writing guidance supports modelling, practice and reflection as a core instructional cycle. A useful model is therefore not a trophy text. It is teaching material that is inspected, discussed, imitated selectively and then left behind.
Do not begin with admiration
“This is a great essay” tells the learner little. Ask what makes it effective for this task, audience and purpose.
What job does the introduction perform? Where is the thesis? How does the paragraph develop its idea? Where is evidence integrated? What has the writer deliberately omitted?
Annotate decisions, not adjectives
A weak model lesson labels writing “interesting”, “descriptive” or “strong”. A stronger lesson labels moves: establishes context, narrows question, signals contrast, embeds evidence, qualifies claim, delays detail, returns to thesis.
Decision labels transfer more easily than praise words.
Compare two models when possible
One model can accidentally look like the only correct form. Two successful models reveal variation.
Compare two introductions that both work for the same broad task but use different routes. Ask what function they share despite surface differences.
Use near-miss models too
A text that is almost successful can be more instructive than a flawless exemplar. Ask where it begins to drift, over-explain, under-develop or lose audience fit.
Students often learn boundaries by comparing good and near-good.
Start with one move
Do not deconstruct every feature in a sophisticated model at once. If the lesson target is paragraph development, ignore elegant vocabulary unless it directly serves that target.
Narrow attention reduces cognitive overload and makes imitation purposeful.
Sentence imitation should preserve structure, not content
Take a useful sentence pattern: “Although X, Y because Z.” Ask the learner to use the same relationship on a different topic.
The learner is borrowing architecture, not wording. The new sentence should be semantically independent.
Paragraph imitation should preserve job sequence
A model paragraph may move claim → explanation → evidence → interpretation → qualification. Ask the learner to write a new paragraph with the same job sequence on another topic.
Then vary the order where appropriate so the structure does not become a cage.
Narrative models teach information control
A strong story may delay a fact, shift pace, reveal character through action or use dialogue to expose conflict. Identify the move, then change character, setting and event.
Creative transfer requires a new surface with the same craft principle.
Model compositions should not become phrase banks only
Students often copy striking phrases because they are memorable. This can produce writing that sounds polished but does not fit the new context.
Ask what the phrase did in the original. If its function can be reproduced more naturally with different language, that is stronger learning.
Imitation is a stage, not the destination
Early imitation reduces decision load. The learner can focus on one pattern while other choices are temporarily constrained.
But support must fade. After one or two close imitations, change topic, audience, genre or structure so the learner has to select the move independently.
Variation tests understanding
Change one variable at a time. Same structure, new topic. Same topic, new audience. Same argument, new evidence. Same scene, new viewpoint.
If the student can adapt successfully, the underlying decision is becoming portable.
Reverse engineering makes craft visible
Give the finished model and ask students to infer the likely plan behind it. What paragraph jobs must the writer have chosen? What evidence did they need? Which decision probably came before drafting?
This turns the text into a trace of process.
Remove a section and reconstruct it
Hide the model’s conclusion or one body paragraph. Ask the student to write what should come next based on the existing architecture. Then compare with the original.
The purpose is not to match wording. It is to test whether the learner understands the text’s direction.
Change the audience
Take a model explanation written for adults and rewrite it for younger learners. Which vocabulary, examples and sentence structures change? Which core content remains?
This reveals function beneath wording.
Change the genre
Turn an explanatory paragraph into a short speech or infographic. The learner has to decide which relationships survive and which need new representation.
Genre transfer is a strong test of understanding.
Use model–student comparison carefully
Do not simply ask, “Why isn’t yours like this?” Compare one target feature at a time. Perhaps both texts have clear claims, but the model develops evidence more fully.
Specific comparison preserves agency and avoids treating the exemplar as an identity standard.
Use anonymised student models where appropriate
Peer-level models can show that strong work is attainable. They also contain realistic strengths and weaknesses that invite analysis.
Protect privacy and obtain appropriate permission when using real student work.
Teacher-written models can expose process
A tutor can deliberately write a model in front of students, pausing to explain choices. This is especially powerful because learners see drafting rather than only the polished endpoint.
Show uncertainty and revision. Expert writing is not effortless.
Think-aloud modelling should be brief
Explain one decision: “I am moving this sentence because the evidence arrives before the reader knows the claim.” Then let the learner try.
Too much expert narration makes the learner a spectator.
Primary 1–2: models establish sentence patterns
Use short model sentences and tiny texts. Notice capitals, sentence completeness, sequence and simple description.
Ask the learner to write a new sentence with the same relationship, not simply copy vocabulary.
Primary 3–4: models reveal paragraph jobs
Compare a model paragraph with a weaker one. Ask which sentence gives the point, which develops it and which detail supports it.
Then write a new paragraph on another topic using the same jobs.
Primary 5–6: models teach selective craft
Use compositions, situational writing and comprehension answers to show planning, development, evidence and precision.
Teach students to collect moves rather than memorised sentences.
Secondary 1–2: models become comparative
Students can compare several successful approaches to the same task. Ask what remains invariant: relevance, coherence, evidence, audience fit, clarity.
This protects against formula dependency.
Secondary 3–4: models should survive unseen tasks
After studying an exemplar, give a prompt that looks different but requires a related decision. Do not announce which feature to transfer.
Independent selection is the test.
JC: models reveal argument architecture
Advanced students can study thesis calibration, paragraph sequencing, evidence integration, counterargument and qualification.
They should also critique models. A strong essay is not beyond evaluation.
Model answers in comprehension need care
A model answer shows scope and expression, but students may memorise wording instead of reasoning. Hide the model, return to the question and ask how the answer was constructed from evidence.
Then change the passage.
Model oral answers should be flexible
Students often imitate fluent spoken responses too literally. Focus on structure: direct answer, reason, example, connection. Then change topic and follow-up question.
Spoken language needs adaptability more than memorised polish.
Use model text stripping
Remove decorative vocabulary while preserving structure. Does the text still work? If yes, the architecture may be carrying more of the quality than the “good words”.
This is a useful antidote to vocabulary worship.
Use model text expansion
Take a concise model and ask where additional detail would genuinely improve it. Students learn that good writing can be developed without becoming bloated.
Use model text compression
Cut a model paragraph to half its length without losing the main reasoning. This reveals which components are essential and which are optional.
Use sentence swapping
Move two sentences in a model. What breaks? Cohesion and information order become visible when the text is deliberately damaged.
Use paragraph swapping
Reorder body paragraphs. Does the line of argument weaken? Ask why the original sequence may have been chosen.
Use imitation with constraints
Borrow the structural move but prohibit copying any content word longer than four letters from the model. The arbitrary constraint forces the learner to reconstruct the move rather than reproduce the language.
Use anti-models
Write a deliberately overloaded introduction, unsupported paragraph or dialogue-heavy story. Ask students to repair it using principles learned from good models.
Error repair deepens understanding of the standard.
Model texts should include diversity of voice
If every exemplar sounds identical, students may mistake one stylistic preference for quality itself. Use different voices, genres and approaches where they all meet the task successfully.
Do not imitate a writer’s identity
The goal is not to make students sound like a famous author, teacher or top student. Teach transferable craft moves such as contrast, scene control, evidence integration or qualification.
Originality grows when principles are recombined through the learner’s own material.
Teach attribution when borrowing ideas or wording
Model study should also teach ethical boundaries. Copying a sentence or idea into assessed work may require quotation or citation depending on context.
Distinguish learning imitation from presenting someone else’s language as original work.
Use models before drafting and after drafting differently
Before drafting, a model can provide architecture. After drafting, a model can become a comparison tool for one target feature.
Do not replace the learner’s draft with the model’s solution.
Use models to generate criteria
Ask students what several strong examples have in common. Turn those observations into a short checklist for the current task.
Criteria become more meaningful when learners can see where they came from.
Then test the criteria on a new text
A useful criterion should help judge unfamiliar work. If it only describes one model’s surface features, it is too narrow.
Three learners, three model-text problems
Maren refuses models because she wants to be original, so she misses useful structural lessons. Iona copies models too closely and produces polished but generic work. Leonie notices many techniques but cannot decide which one matters for the new task.
Maren deconstructs function without copying style. Iona uses variation constraints. Leonie selects one target based on the new prompt’s actual need.
A 90-minute model-text tutorial
- 10 minutes: define the target skill.
- 15 minutes: read one or two models for that target.
- 15 minutes: annotate decisions and compare alternatives.
- 15 minutes: imitate one move on a changed topic.
- 15 minutes: vary audience, genre or structure.
- 15 minutes: independent task with no model visible.
- 5 minutes: compare and identify what transferred.
Measure transfer, not resemblance
A student can produce a text that looks very different from the model and still demonstrate successful learning if the underlying decision has transferred.
Similarity is not the goal. Functional improvement is.
What progress looks like
The learner increasingly identifies why a model works, names transferable decisions, imitates one move without copying language, varies that move under changed conditions, critiques exemplars and eventually selects useful strategies without needing a model open beside the task.
The model has done its job when the learner no longer needs it.
Where this tutorial connects inside SETC
- English Writing and Composition
- Creative Writing Handbook
- How Feedback Changes Performance
- English Tutorial Lesson No.0003
- English Tutorial Lesson No.0011
- Primary English
- Secondary English
- Advanced English / General Paper
External research and teaching references
- Institute of Education Sciences / What Works Clearinghouse — Teaching Elementary School Students to Be Effective Writers
- British Council TeachingEnglish — Teaching Resources
- Cambridge English — Activities for Learners
Final tutorial challenge
Choose one short model and identify exactly one transferable move. Hide the model. Reproduce the move on a completely different topic. Then change the audience and revise it again. Finally, explain what stayed constant across all three versions.
If the learner can name the principle without reproducing the original wording, imitation has become transfer.
English Learning Library · Primary English · Secondary English · Advanced English / General Paper
Continue with the English Tutorial collection.


Leave a Reply