HookSteal the machinery, never the cargo
On 23 September 2019, Greta Thunberg stood at the United Nations in New York and delivered roughly five hundred words that travelled around the world. Four times she loaded the same three: 'how dare you'. The machinery is visible once you stop being moved long enough to look — a pronoun war in which 'you' takes the blame and 'we' takes the consequences; an accusatory second person hammered home by repetition at the head of successive sentences; statistics read out like charges on an indictment; the childlike concrete ('my dreams and my childhood') set against the bureaucratic abstract ('empty words'). If you chose that speech as a style model, you would not copy its anger. You would strip it for parts — the pronoun system, the repetition engine, the concrete-against-abstract collisions — and rebuild the machine to carry your own cargo.
That is the entire original-writing task in one move. The second half of the Language in Action folder asks for one 750-word piece from three routes — the power of persuasion, the power of storytelling, or the power of information — plus an annotated style model and a 750-word commentary. The piece itself is where AO5 lives: this is one of only two places in the whole qualification where you are assessed on expertise and creativity as a writer, and the only one with unlimited drafting time. The commentary earns its credit differently — AO3 for the evaluated relationship between your choices, your style model and your context, and AO2 for the concepts underneath those choices. Students sink the folder by treating its three documents as separate homeworks. They are one engineered object.
ModelThree powers, three crafts — choosing your route
Persuasion — a speech, an opinion editorial, a campaign appeal — is argument architecture: a stance, an audience positioned by pronouns, patterning that makes sentences quotable (triads, anaphora, antithesis), and evidence sequenced for feeling as well as logic. Storytelling — short fiction, a dramatic monologue, the opening of something longer — is the hardest craft to control: a narrative voice held absolutely steady, focalisation that stays put, scene rendered rather than summarised, dialogue that characterises instead of merely informing. At 750 words you render one situation completely; you do not compress a saga. Information — a feature, an explainer, a piece of popular science — is the invisible craft: sequencing from what the reader knows to what they don't, analogy as load-bearing structure, definitions that never condescend, a register that welcomes.
Choose the genre you already read fluently. Your ear is trained wherever your reading hours actually went, and a moderator can hear the difference between a genre inhabited and a genre imitated within a paragraph. Ambition matters too — a controlled risk outscores a safe flatline — but ambition in a genre you don't read is just risk.
ModelThe style model — reverse-engineer, then rebuild
A style model is a real published text in your target genre, chosen before you draft and annotated across the language levels: how it opens, how its sentences breathe, which lexical fields it runs, how it handles its reader (address, presupposition, politeness), how it closes. What you transfer is machinery, never phrasing. 'Cold open on one named individual; second-person turn; statistical escalation; circular close' is a transferable architecture. Its actual sentences are plagiarism waiting to be found — and the annotated model travels in the folder precisely so the moderator can see the relationship between its techniques and yours. That visible relationship is where the AO3 credit flows from.
Suppose your route is persuasion and your style model is a food-bank fundraising email that opens: 'It's 4pm on a Tuesday, and Maria is deciding which of her three children eats breakfast tomorrow. Not whether. Which.' The annotation writes itself: present tense for witnessed immediacy; one named individual, because a single story resists the numbness of statistics; deictic grounding (4pm, Tuesday) that plants the reader inside the scene; and a minor-sentence pivot — 'Not whether. Which.' — using end-focus to land the unbearable word last.
Now the transfer. Your piece campaigns against a youth-club closure: 'It's 9pm on a Friday, and fifteen-year-old Danny is choosing between the bus shelter and the precinct. The youth club used to make that choice for him. The council unmade it in March.' Same machinery — present tense, named individual, deixis, the short-sentence turn — carrying entirely new cargo. Spelt out in the commentary, that transfer is exactly what the moderator is looking for.
MechanismDrafting for AO5 — control is the creativity
AO5's two nouns are expertise and creativity, and in a 750-word piece both cash out as control: register held consistent to the final line; rhythm varied on purpose, so the long cumulative sentence buys the three-word one that follows it; lexis precise rather than ornamental; a structure with a shape — circular echo, sectional turn, delayed reveal — rather than paragraphs that merely stop. Hunt clichés without mercy: the first phrase that arrives is everyone's phrase, which is how it arrived so fast.
Then exploit the NEA's one luxury the exam room never offers: time. Keep every draft and date it. Log what you changed and why — 'moved the question to its own paragraph; buried mid-paragraph it read as genuine enquiry' — because that log becomes the spine of the commentary. Read the piece aloud before calling it finished; the ear catches register wobble the eye forgives. And respect 750 words as a discipline rather than a suggestion: in a piece that short, every paragraph is re-auditioning for its place, and bloat dilutes exactly the control being rewarded.
MechanismThe commentary — engineering notes, not a diary
The commentary is 750 analytical words marked for AO2 and AO3: the same terminology you would spend on an unseen text, spent on your own. Shape it in three parts. Aims, in one short paragraph: genre, audience, purpose, and a named publication context. Then the body as a series of engineering notes, each one welding four things — my choice, the style model's licence for it, the intended pragmatic effect, and an evaluation of whether it survived redrafting. Then a close on what the finished piece actually does, not what you hoped it would.
Two modes sink commentaries every year. The diary — 'first I wrote an opening, then I added a rhetorical question…' — narrates chronology instead of analysing choices. And the unearned effect — 'to engage the reader', 'to make it flow' — claims outcomes without mechanisms, which is exactly the analytical crime you have spent two years learning not to commit on other people's texts.
Same feature, two commentaries. Weak: 'I used a rhetorical question at the start to engage the reader.' It names a form, claims an effect no mechanism supports, and mentions neither the model nor the drafts. Strong: 'The opening interrogative — “Whose job is Danny, exactly?” — adapts my style model's habit of indicting its reader directly; in draft one it sat mid-paragraph and read as genuine enquiry, so I isolated it as a one-sentence paragraph, where the surrounding silence turns a question into an accusation.' One sentence, and it carries the feature, the model's licence, the pragmatic mechanism, the drafting decision and the evaluation — AO2 and AO3 in a single breath.
CaseThe folder as the moderator meets it
Your folder reaches a stranger as three documents — the piece, the annotated style model, the commentary — and it has to argue for itself, because you are not in the room. A timetable that works: weeks one and two, choose the route and the model, and annotate it properly; week three, first draft; your teacher may respond with general advice, because guidance is allowed but editing is not, and the redrafting decisions must be demonstrably yours; weeks four and five, redraft and keep the log; week six, write the commentary from the log while the decisions are still warm.
The folder-sinkers moderators flag are always the same. Style models chosen after the piece was written, whose 'influence' reads as generic coincidence. Pieces that swelled to 900 words because the writing was going well, diluting the very control AO5 pays for. Commentaries that describe instead of evaluate. Avoid those three and you are already in the upper half — because this is the one part of the A-level you can hand in already finished, polished, and read aloud twice.
VocabularyKey terms the mark scheme pays for
TrapsMisconceptions that cost marks
ExamWhat examiners want
Know where each mark lives. The piece itself is AO5 territory: expertise and creativity demonstrated as control of genre, register and structure across a disciplined 750 words. The commentary is marked for AO3 — the evaluated relationship between your writing, its style model and its context — and AO2, the linguistic concepts your choices put into practice. The annotated style model is what makes both visible: choose it first, annotate it at the language levels, and let the commentary keep pointing back to it.
Name a publication context in your aims — 'a comment piece for an online broadsheet', 'a charity appeal email to lapsed donors' — because a real context makes every register decision checkable rather than a matter of taste. Keep dated drafts: the folder is teacher-marked and AQA-moderated, general guidance is permitted but editing is not, and the drafts are your evidence that the decisions were yours.
Then exploit the format. This is one of only two AO5 assessments in the qualification, and the only one without a clock — the other is Paper 2's 45-minute directed writing. A piece read aloud, redrafted twice and trimmed back to 750 will beat a first draft of genius almost every time, and the commentary you write from a drafting log sounds like an engineer, which is precisely the voice the top band is listening for.