Subject playbook

AI for English Teachers

Use AI well in English — for modelling, differentiation, feedback and resources — without ever trusting a fabricated quote or a thesis-light reading.

Modules

Each module: English learning outcomes → short, accurate subject input → a hands-on activity using the Lessio generator on real English work → interactive knowledge checks. Mapped to the Australian Professional Standards for Teachers.

Click a module to read it.

1. 1

   Where AI actually helps in English — and where it fails

   The high-value English use cases worth your time, set squarely against the four ways AI fails for English specifically: fabricated quotes, invented techniques, thesis-light analysis, and the essay as the integrity front line.~45 min

   By the end of this module you'll be able to:

   • Name the English tasks where AI genuinely saves time — and route the right job to the right tool.
   • Identify the four English-specific failure modes and explain why a fluent English answer is so often a wrong one.
   • Decide, for any English task, what AI may draft and what your reading must own.

   Standards2.1 Content and teaching strategies of the teaching area2.5 Literacy and numeracy strategies3.4 Select and use resources

   Assumed knowledge — straight to English

   You've done the flagship course, so you already hold teacher-in-the-loop, the policy stack, the student-data hard line, and RICE (Role · Intent · Constraints · Examples). This playbook does one thing: it makes those count in an English classroom. We won't relitigate the ethics — we'll go subject-deep.

   Where AI genuinely helps an English teacher

   These are the use cases worth your time. In every one, AI drafts and you read, mark and decide.

   • Model and exemplar texts at a target standard. A "Band 4" body paragraph, a sophisticated thesis statement, a deliberately mid-range response to mark up with a class. Exemplars are gold in English and slow to write from scratch.
   • Differentiating a text to a reading level, or for EAL/D learners — a plain-English version that keeps the ideas, with a glossary of the vocabulary you simplified.
   • Comprehension question sets across literal, inferential and evaluative levels for a given text or extract.
   • Vocabulary and sentence-level scaffolds for extended writing — tier-two and tier-three word banks, sentence starters, paragraph frames (e.g. point–evidence–analysis).
   • Analysing the language forms and features of a text as a starting list to test against the words on the page.
   • Writing, discussion and creative stimulus — provocations, "what if" prompts, image-and-line pairings, debate positions.
   • Rubric-to-student-friendly criteria — turning marking guidelines into "I can…" success criteria.
   • Feedback frames on extended responses — a strength, a priority, a next step against your criteria, for you to make specific and kind.

   Where it fails — and these are English's failures, not generic ones

   English breaks AI in four particular ways. Learn them by name; the rest of this playbook is built around them.

   Failure	What it looks like	Your move
   Fabricated quotations	It "quotes" a line that is not in the text, or warps the real wording, and cites act/scene/page numbers that don't exist.	Never trust an AI-supplied quote. Check every one against the actual text before it reaches a student or a marking guide.
   Invented or misattributed techniques	It labels a line "juxtaposition" when there's none, or names a device the passage doesn't use.	Verify every technique claim against the words on the page; if you can't point to it, cut it.
   Thesis-light, generic analysis	Fluent paragraphs that float — feature-spotting with no sustained argument, no conceptual line, no so what.	Treat AI analysis as raw material, never a model of thinking. Demand and supply the argument yourself.
   The essay = the #1 integrity risk	Extended responses are the easiest task in the school to outsource to AI.	Design for authorship (Module 4) — process, drafts, in-class checkpoints — not detection.

   The throughline: in English, plausible prose is the trap. AI is most fluent exactly where it is least trustworthy — quotation, technique and argument. Fluency is not evidence.

   Why English is different

   Maths has a right answer the engine can be grounded to. English runs on a primary text the model has only read about — so it confabulates the specifics (the quote, the device) while sounding completely assured. That is why the single most important English-AI habit is return to the text: the poem, the novel, the film, the extract in front of you is the authority — not the model.

   Activity — map the helps and the traps (10 min)

   Open the Lessio generator and produce a short model analytical paragraph on a text you teach (give it the text and a target standard). Then audit it like a marker: (1) highlight every quotation and check it against the actual text; (2) circle every named technique and confirm it's really there; (3) judge the thesis — is there a sustained argument, or just feature-spotting? Write one sentence on what AI saved you and one on what your reading had to fix. That gap is the job this playbook trains.

   Knowledge check

   1Name the four English-specific ways AI fails.

   Show answer

   2AI hands you a perfectly fluent paragraph on a poem. Why is fluency not enough in English?

   Show answer

   3Give three English tasks where AI genuinely saves a teacher time.

   Show answer

   Mark this module complete
2. 2

   Prompt craft for English — grounding RICE in the syllabus

   RICE, but English-grade: anchoring prompts in the real NSW English outcomes, the subject-specific guardrails that stop fabricated quotes and thin analysis, and a ready English prompt library.~50 min

   By the end of this module you'll be able to:

   • Anchor a RICE prompt to the relevant NSW English focus area and outcomes for the stage.
   • Build the English guardrails — 'use only the text I paste', 'quote exactly', 'no invented techniques' — into the prompt itself.
   • Iterate an English output toward a sustained argument, and bake in a self-check that surfaces fabricated quotes.

   Standards2.1 Content and teaching strategies of the teaching area2.5 Literacy and numeracy strategies6.2 Engage in professional learning

   RICE, taken to English

   You know RICE from the flagship. English just demands more of every letter — because the failure modes from Module 1 have to be designed out in the prompt.

   • R — Role & context. "You are an experienced NSW Stage 5 English teacher planning for a mixed-ability Year 10 class with several EAL/D learners."
   • I — Intent + the syllabus anchor. Tie the task to the focus area and outcomes. Stage 4 English has three: Reading, viewing and listening to texts (EN4-RVL-01), Understanding and responding to texts (EN4-URA-01, EN4-URB-01, EN4-URC-01) and Expressing ideas and composing texts (EN4-ECA-01, EN4-ECB-01). Stage 5 mirrors this with Understanding and responding to texts (EN5-URA-01, EN5-URB-01) and Expressing ideas and composing texts (EN5-ECA-01, EN5-ECB-01). Name the one you're targeting.
   • C — Constraints & format, including the English guardrails. This is where you stop the hallucinations: "Use only the extract I paste below — do not draw on any other text. Quote exactly and put quotation marks around every quotation. Do not name a technique unless it is demonstrably in the extract. Australian spelling."
   • E — Examples & evaluation. Show the standard you want, and force a self-check: "Match the register of this Band 5 sample. At the end, list every quotation you used so I can verify each against the text, and flag any technique you were unsure of."

   The English guardrails — say these every time

   Because of Module 1, three constraints belong in almost every English prompt:

   1. "Use only the text I provide" — closes the door on the model importing half-remembered lines from its training.
   2. "Quote exactly; mark every quotation" — makes verification fast and forces honesty about wording.
   3. "Don't invent techniques — only name what's in the text" — kills the misattributed-device problem.

   And one self-check: "List your quotations and flag anything unverified" — which turns the model's worst habit into your checklist.

   English-specific pitfalls in prompting

   • Paste the text; never assume the model 'knows' it. Even canonical texts come back with warped quotes. Ground it on your extract.
   • Push past feature-spotting. A first draft will list devices. Iterate: "Now connect these to a single argument about how the composer positions the responder — make it conceptual, not a list."
   • Mind register and audience. "Year 7 reading age" and "Band 6 marker" are different worlds — constrain explicitly.
   • Respect ICIP. For texts by and about Aboriginal and Torres Strait Islander peoples, do not have AI fabricate cultural content, Country, or community perspectives — source those properly and keep AI to mechanical scaffolding.

   See the difference

   Weak English prompt	Strong English prompt
   "Analyse this poem."	"You are a NSW Stage 6 English Advanced teacher. Using ONLY the poem I paste below, write a Band 5 analytical paragraph on how the poet represents memory. Quote exactly and mark every quotation; do not name a technique unless it is in the poem. Sustain one argument — don't list features. Then list each quotation you used so I can verify it."
   "Give me comprehension questions."	"You are a NSW Stage 4 English teacher (focus: Reading, viewing and listening, EN4-RVL-01). From the extract below, write 8 comprehension questions for a mixed-ability Year 8 class: 3 literal, 3 inferential, 2 evaluative, with an answer key drawn only from the extract. Flag any question whose answer isn't explicit in the text."

   Activity — build and harden an English prompt (15 min)

   Take a text you teach next term. Write your instinctive one-line prompt, then rebuild it with RICE, the syllabus anchor (name the focus area), and the three guardrails plus the self-check. Run it, then verify every quotation it returns against the text. Now do the same job in Lessio and notice how much of the syllabus-anchoring is already handled — your attention is freed to police quotation and argument.

   Knowledge check

   1Which three guardrail instructions belong in almost every English prompt?

   Show answer

   2You're prompting for Stage 5 'Understanding and responding to texts'. Which outcome codes anchor it?

   Show answer

   3A first AI analysis comes back as a list of techniques. What's the iteration move?

   Show answer

   Mark this module complete
3. 3

   Planning & resources for English with Lessio

   Using the grounded generator for English: a Stage 5 reading unit, a differentiated resource, and a connected scope & sequence — drafted to the verbatim NSW English syllabus, then shaped by you.~55 min

   By the end of this module you'll be able to:

   • Generate a syllabus-aligned English scope & sequence, unit and resource as one connected set in Lessio.
   • Verify English outcome codes, text requirements and coverage before a document is used.
   • Differentiate an English resource for reading level and EAL/D while holding the same outcome.

   Standards2.2 Content selection and organisation2.3 Curriculum, assessment and reporting3.4 Select and use resources

   Why a grounded engine matters more in English

   A general chatbot writes an English program from general text about teaching English. Lessio drafts to the NSW DoE program template and grounds every draft in the verbatim NESA English syllabus for the focus area you choose — so a Stage 5 unit reads like your faculty's work, with the real outcomes attached. It is still a draft: you set the text choices, the sequence, the conceptual line and the assessment. But you start from syllabus-accurate scaffolding instead of a blank page.

   The DoE template Lessio follows is the five-column teaching-and-learning sequence — Outcomes & content · Teaching and learning activities · Evidence of learning · Differentiation & adjustments · Registration & evaluation — wrapped around an outcomes list, a needs analysis under the UDL headings, and a reflection block. In English that scaffolding holds your reading–responding–composing cycle together.

   English covers more than novels — and the syllabus says so

   Across stages, NSW English works with spoken, print, visual, media, multimedia and digital texts, including quality Australian texts, texts by and about Aboriginal and Torres Strait Islander peoples, and a range of perspectives. When you generate, name the text type you want (a film, a speech, a multimodal text, poetry) so the unit reflects the real breadth of the course — and remember the ICIP line: source Indigenous texts and perspectives properly; never have AI invent cultural content.

   Three English jobs Lessio does well

   1. A Stage 5 reading unit. Target Understanding and responding to texts (EN5-URA-01, EN5-URB-01). Lessio drafts the unit to those outcomes with a reading–responding focus; you choose the core text, build the conceptual question, and sequence the close-reading.

   2. A differentiated resource. From one comprehension or analysis resource, generate an enable version (sentence starters, a glossary, a PEA frame, a plain-English extract) and an extend version (a harder lens, an open-ended composing task) — both targeting the same outcome. This is UDL in practice and, for students with disability, a legal expectation under the Disability Standards for Education 2005 — you confirm each adjustment still hits the outcome.

   3. A connected scope & sequence. Planning is a chain: scope & sequence → unit/program → resources → assessment, all pointing at the same outcomes. Generate it connected — e.g. a Stage 4 year that moves across Reading, viewing and listening (EN4-RVL-01) into Expressing ideas and composing texts (EN4-ECA-01, EN4-ECB-01) — then shape it. In Stage 6, anchor to the course module: English Standard and Advanced both teach the Common Module Texts and Human Experiences (Year 12 — Standard EST-12-01 onwards; Advanced EAV-12-01 onwards), with English EAL/D running the parallel module (EEA-12-01 onwards).

   Review-before-use — the English edition

   A 90-second discipline before any draft becomes a real document:

   1. Codes — every outcome code matches the current English syllabus and the right focus area.
   2. Text requirements — the unit honours the stage's text breadth (visual/media/multimodal as well as print) and any Australian / Aboriginal and Torres Strait Islander text requirements, sourced respectfully.
   3. Quotes & references — any quotation, page or act/scene reference in a resource is verified against the actual text (Module 1).
   4. Sequence — the reading–responding–composing build is sound for your students.
   5. Inclusion — language, texts and examples are inclusive and culturally safe.

   If you couldn't defend it in a faculty programming meeting or a registration audit, it isn't ready.

   Activity — generate a Stage 5 reading unit, then make it yours (15 min)

   In Lessio, generate a Stage 5 reading unit anchored to Understanding and responding to texts (EN5-URA-01, EN5-URB-01) for a core text you teach. Then: verify the two outcome codes against the syllabus, swap in your text and conceptual question, generate one differentiated version of a resource in it, and check every quotation the resource contains. The edits you make are the visible proof of your professional judgement.

   Knowledge check

   1What does Lessio ground an English unit in that a general chatbot can't reliably reproduce?

   Show answer

   2Beyond print novels, what text types must an English unit reflect — and what's the ICIP caution?

   Show answer

   3You generate an 'enable' version of a Stage 5 analysis resource. What must you confirm?

   Show answer

   Mark this module complete
4. 4

   Assessment, feedback & integrity in English

   Essays and extended responses: building feedback frames, writing assessment notifications and marking guidelines with Lessio, and the authorship front line — assuring it by design, not by 'detectors'.~50 min

   By the end of this module you'll be able to:

   • Use AI to draft feedback frames on extended responses, then make them accurate, specific and kind.
   • Generate an English assessment notification with marking guidelines in Lessio, and check it for validity and fairness.
   • Assure authorship of essays the way NESA expects — by task design and process, not detection software.

   Standards2.3 Curriculum, assessment and reporting5.1 Assess student learning5.2 Provide feedback to students on their learning

   English is the authorship front line — own it

   Extended responses are the most outsourceable task in the school, and AI writes plausible essays effortlessly. So in English, integrity isn't a footnote — it's a design problem you solve up front. From the flagship you know NESA's stance: schools decide whether AI is permitted task by task, must uphold HSC and RoSA integrity, and treat malpractice — plagiarism, collusion, or presenting AI work as a student's own — as a serious matter. NESA's guidance is to assure authorship by design, not by 'AI detectors' (which are unreliable).

   Assure authorship by design — what that looks like in English

   • Make the process visible. Staged drafts, planning, annotated bibliographies, in-class writing checkpoints — assess the journey, not just the final essay.
   • Talk to the writing. Brief viva-style conversations ("walk me through your thesis", "why this quote?") surface genuine authorship fast and fairly.
   • Anchor to a studied text and a specific question. A tightly text-specific prompt, with quotations the student must integrate from class work, is far harder to outsource than a generic theme.
   • Be explicit about what's permitted. State on the notification whether AI may be used, and how it must be disclosed — model the honesty you ask of students.

   AI 'detectors' produce false positives and false negatives and can't be defended in a malpractice process. Replace the catch-them mindset with a design-it-out one.

   Feedback — frames from AI, judgement from you

   AI is genuinely useful for structuring feedback on extended responses at speed; it does not know your student or your marking standard. Use it for a frame — one strength, one priority, one specific next step against your criteria — then make it accurate, specific and kind for the actual writer. Two English-specific cautions:

   • De-identify first. Never paste an identifiable student's response into a general tool — strip the name (and anything identifying), or use your school's approved, secured environment.
   • Police the quotes in the feedback too. If the frame references the student's quotations or technique use, verify them against the response — don't let the model invent what the student 'did'.

   Building English assessment with Lessio

   Lessio drafts to the syllabus, which is exactly what an assessment needs.

   • Assessment notification — generate a Stage 5 notification anchored to Understanding and responding to texts or Expressing ideas and composing texts, with the task, the outcomes, the conditions, and a clear statement on permitted AI use.
   • Marking guidelines — generate criteria or a marking guideline grid aligned to the outcomes, then rewrite it as student-friendly success criteria ('I can…') so students can self-assess.
   • Validity & fairness check — confirm the task measures the outcome (an analytical-response task should assess Understanding and responding, not neat handwriting), that conditions are fair, and that reasonable adjustments (Disability Standards for Education 2005) are built in.

   Review-before-use — assessment edition

   Before any English assessment goes live: outcomes correctly cited; the task is valid (measures what it claims) and reliable (the marking guideline discriminates fairly); quotations and stimulus texts verified; AI-use rules stated; and adjustments for EAL/D and disability included.

   Activity — notification + marking guideline, then feedback frame (15 min)

   In Lessio, generate a Stage 5 analytical-response assessment notification with marking guidelines, anchored to EN5-URA-01 / EN5-URB-01. Check it for validity (does it measure understanding-and-responding?) and add a line stating whether AI is permitted and how it's disclosed. Then take a de-identified student response and use AI to draft a feedback frame against your criteria — and verify every quote it attributes to the student before you'd ever hand it back.

   Knowledge check

   1Why does NESA steer schools away from 'AI detectors' for essays, and what's the alternative?

   Show answer

   2What two English-specific checks apply when AI drafts a feedback frame on a student response?

   Show answer

   3You generate an analytical-response notification in Lessio. What makes it 'valid'?

   Show answer

   Mark this module complete
5. 5

   Capstone — build and critique a real English unit

   Build a connected English unit, resource and assessment with Lessio, then critique it as an English specialist — verifying every quote and technique, the argument, and the syllabus fit — and log it as PD.~50 min

   By the end of this module you'll be able to:

   • Build a connected English unit, resource and assessment with Lessio, end to end.
   • Critique the set as an English specialist — verifying quotes, techniques, argument quality, text breadth and outcome fit.
   • Self-assess against the Ethical-Use Checklist, reflect, and record the hours as Standards-relevant PD.

   Standards2.1 Content and teaching strategies of the teaching area5.1 Assess student learning6.2 Engage in professional learning

   The task — a real, connected, defensible English set

   Choose a text and stage you'll teach next term. Using Lessio, build and then critique:

   1. A unit / program for one English focus area (e.g. Stage 5 Understanding and responding to texts, EN5-URA-01 / EN5-URB-01, or a Stage 4 reading–composing unit, EN4-RVL-01 → EN4-ECA-01/ECB-01).
   2. A resource within it — a model paragraph, a comprehension set, or a differentiated text — with an enable and an extend version.
   3. An assessment notification with marking guidelines for the unit, with permitted-AI-use stated.

   Critique it as an English specialist — the subject lens

   This is where your expertise shows. Run every artefact through the English checks:

   • Quotation audit — every quotation in every resource and marking guide verified against the actual text. (Zero tolerance: one fabricated quote fails the artefact.)
   • Technique audit — every named device confirmed present in the text; cut what isn't.
   • Argument quality — any model response sustains a real argument, not a list of features; the marking guideline rewards thinking, not feature-spotting.
   • Text breadth & ICIP — the unit honours the stage's range of text types and any Australian / Aboriginal and Torres Strait Islander text requirements, sourced respectfully.
   • Outcome fit & validity — codes are current and correct; the assessment measures the outcome; adjustments for EAL/D and disability are built in.

   What good looks like

   A connected, syllabus-accurate English set you'd actually teach — drafted by AI, unmistakably read, corrected and owned by you. Your edits — the verified quotes, the sharpened thesis, the chosen text, the fair marking guideline — are the evidence of your professional judgement and exactly what teacher-in-the-loop looks like in English.

   Self-assessment — the Ethical-Use Checklist

   Run your capstone against all five checklist items on this page, especially the line on verifying every AI-supplied quote and technique against the actual text. Every box should be honestly tickable. If one isn't, fix the artefact — that is the learning.

   Reflection — write a short response

   • What did AI genuinely save you time on in English, and what did you have to correct?
   • Where did your reading of the text override the draft — a wrong quote, a phantom technique, a thin argument?
   • How did you assure authorship in the assessment you built?
   • One rule you'll keep for using AI responsibly in your English classroom.

   Log it as professional learning

   This module is your assessment: a complete, critiqued English set plus your ethical-use reflection — keep it as evidence of practice. Since NESA removed the Accredited/Elective PD distinction in 2024, Standards-relevant learning like this counts toward your 100 maintenance hours — record it in your eTAMS PD log against the Standards it addresses (especially 2 — knowing English content and how to teach it — plus 5, 6 and 7). Your English faculty can also run this playbook as part of its professional-learning plan or KLA team time.

   Activity — build, then critique (15 min)

   Generate your connected English set in Lessio, then run the full English critique above on it — auditing every quotation and technique, judging the argument, and confirming outcome fit. Capture your three biggest corrections; they are the proof of your professional judgement.

   Knowledge check

   1What single check has zero tolerance when critiquing an AI-built English set?

   Show answer

   2What turns an AI-generated English unit into defensible professional work?

   Show answer

   3How does this playbook count toward your NESA professional-development hours?

   Show answer

   Mark this module complete

Take-away prompt library

Ready, RICE-shaped prompts for common NSW teaching jobs (Module 3). De-identified — copy one, swap in your details, and use it today.

You are a NSW [stage] English teacher. Using ONLY the text I paste below, write a model analytical paragraph at a [target Band/standard] standard on [aspect, e.g. how the composer represents power]. Quote exactly and put quotation marks around every quotation; do not name a technique unless it is demonstrably in the text; sustain one argument rather than listing features. Australian spelling.
Self-check: list every quotation you used and flag any technique you were unsure of so I can verify each against the text.

[paste text]

You are a NSW English teacher supporting EAL/D and below-level readers. Rewrite the extract below in plain English at roughly a [target] reading level, keeping all key ideas and the core subject vocabulary, and add a short glossary of any term you simplified. Use short sentences and active voice. Keep it suitable for the SAME outcome as the original.
Self-check: list any nuance or quotation you altered so I can decide whether to restore it — and confirm you invented nothing not in the extract.

[paste extract]

You are a NSW [stage] English teacher. Using these success criteria [paste], write a feedback frame for the DE-IDENTIFIED student response below: one genuine strength, one priority for improvement, and one specific next step, addressed to the student — warm, specific, jargon-free. Comment only on what is actually in the response.
Self-check: quote back any line you reference and confirm it appears verbatim in the response; do not invent quotes, techniques or content the student didn't write.

[paste de-identified response]

You are a NSW [stage] English teacher (focus: [e.g. Reading, viewing and listening, EN4-RVL-01]). From the extract below, write [N] comprehension questions for a [de-identified class]: a balance of literal, inferential and evaluative, with an answer key drawn ONLY from the extract.
Self-check: flag any question whose answer is not explicit in the text, and confirm every answer is grounded in the extract (no outside knowledge).

[paste extract]

You are a NSW [stage] English teacher. For a [de-identified class] writing about [text/topic], produce: a tier-2 and tier-3 vocabulary bank with brief meanings; five analytical sentence starters; and a point-evidence-analysis paragraph frame. Keep everything aligned to [outcome, e.g. EN5-ECA-01/ECB-01] and to the text I'm teaching.
Self-check: do not supply any quotation as 'evidence' — leave evidence slots blank for students to fill from the actual text, and flag anything I should adapt.

You are a NSW [stage] English teacher (focus: Expressing ideas and composing texts, [e.g. EN4-ECA-01]). Generate three creative-writing stimuli for a [de-identified class] exploring [concept/theme]: each with an evocative opening line or image prompt, a clear link to the composing outcome, and one craft technique students could try. Australian contexts where possible; keep them inclusive and culturally safe.
Self-check: do not attribute any line or technique to a real author or text unless I've provided it — and flag any stimulus that could touch on sensitive content so I can adjust.

Standards-relevant professional learning, mapped to the APST · content verified against national and NSW frameworks, June 2026 · self-log the hours in eTAMS.