GRE Issue Essay Coach (Analytical Writing)
The Issue task gives 30 minutes to take a position on a general-interest claim and defend it with reasons and examples. Since the Sept 2023 shortened GRE it is the only essay (the "Analyze an Argument" task was removed). It is scored 0–6 in half-points by a human rater plus ETS's automated scorer; what separates 4 from 5 from 6 is depth of position and quality of development, not vocabulary display.
Use this when
- The user shares an Issue prompt and wants a plan, a model essay, or a timed-practice setup.
- The user shares their essay and wants a score estimate with rubric-anchored feedback.
- The user asks about structure, examples, counterarguments, or the 30-minute clock.
Do not use this when
- The essay is IELTS — use
ielts-writing-task2 (different rubric, different task logic).
- The user wants their own real argument checked for validity — use
argument-audit.
- The essay is a statement of purpose or application essay — use
apply-sop-writer.
Task facts
- 30 minutes, one prompt, typed in a plain editor (no spellcheck). No minimum length, but scored essays at 5+ are almost always ≥ 500 words because development takes space — length is a byproduct of development, never the goal.
- Every possible prompt is in the published ETS Issue pool (searchable on ets.org) — practice from the real pool.
- Each prompt carries one of six standard instruction stems. The stem changes what a complete response is — ignoring it caps the score:
| Stem asks for |
A complete response must |
| Agree/disagree with a statement |
Take a position AND consider ways the statement might or might not hold |
| Agree/disagree with a recommendation |
Discuss specific circumstances where adopting it would or wouldn't be advantageous |
| Which of two views aligns with yours |
Choose one AND address both views |
| Claim + underlying reason |
Evaluate the claim AND the reason offered for it |
| Consequences of adopting a policy |
Argue from predicted outcomes |
| Statement with a qualification |
Discuss when it holds and when it doesn't |
Score anchors: references/awa-scoring-rubric.md — the 0–6 levels in compact form with the official-source pointer. Consult it before assigning any score so feedback stays anchored to the rubric, not impression.
What actually moves the score
The 4→5→6 gradient, in order of leverage:
- A position with texture. A 4 asserts a side; a 5–6 stakes a claim and marks its limits ("X is generally true, but fails precisely when…"). The instruction stems explicitly invite this — the qualification IS the response to "consider ways it might not hold".
- Developed reasons, few in number. Two or three reasons argued deeply beat five asserted. Each body paragraph: claim → reasoning → specific example → link back to the thesis. "Many historical figures show that…" is a 4; a named case examined in two or three sentences is a 5–6. Examples may come from history, science, current events, literature, or well-drawn hypotheticals — specificity matters, domain doesn't, and invented statistics are unnecessary risk.
- An honest counterargument. One paragraph that states the strongest opposing consideration and answers it (concede scope, then show why the thesis survives). Straw men read as a 4.
- Organization that carries the logic. Paragraph order should be an argument, not a list; transitions signal logical relation ("this holds even when…", "the deeper problem is…") rather than enumeration ("Secondly…").
- Language: varied, precise, and controlled. Errors matter only when frequent enough to obscure meaning; thesaurus decoration lowers precision and reads as a weaker essay. Fix clarity before fanciness.
The 30-minute protocol
| Minutes |
Do |
| 0–4 |
Read the stem twice. Brainstorm both sides (fastest route to a nuanced thesis and a real counterargument). Pick position, jot 2–3 reasons + one example each + the counterargument. |
| 4–26 |
Write: intro (context sentence + thesis with its qualification) → 2–3 reason paragraphs → counterargument paragraph → short conclusion that restates the qualified thesis. Intro and conclusion stay short; the middle is where points live. |
| 26–30 |
Proofread for typos, dropped words, sentence boundaries. No restructuring. |
If time collapses: a complete essay with a conclusion outscores a longer torso that stops mid-thought — always land the ending.
Workflow
Scoring a user's essay
- Identify the prompt's instruction stem and check the essay answered that task — stem-mismatch is the first thing to flag, before any prose feedback.
- Read against the rubric reference and assign a half-point score with a one-line justification per rubric dimension (position, development, organization, language).
- Give feedback in leverage order (the list above): the 1–2 changes worth the most points first, each illustrated by quoting the user's own sentences and showing the upgraded version.
- End with:
Estimated score: X.X / 6
Biggest lever: <one sentence>
Next drill: <one concrete practice instruction>
Calibration honesty: text-only scoring can't see the timed conditions; say the estimate assumes the essay was written in 30 minutes without tools, and ask if it wasn't.
Producing a model essay or plan
- Under timed-realistic constraints: ~500–650 words, achievable structure, no encyclopedic examples the user couldn't produce live.
- Label the parts (thesis, reasons, counterargument) and annotate why each paragraph earns its keep, so the user learns the skeleton, not the specific essay.
Timed practice
Offer prompts from Issue-pool themes (society, education, technology, government, intellectual life), enforce 30 minutes, then score as above. Track scores and recurring feedback across sessions; recurring grammar-level issues in a Vietnamese-L1 user can be drilled with ielts-grammar-coach's general-English mode.
1---2name: gre-issue-essay3description: Draft, score, and improve GRE Analytical Writing "Analyze an Issue" essays against the 0–6 rubric. Use this skill whenever the user shares a GRE Issue prompt or essay ("score my GRE essay", "AWA feedback", "how do I structure the Issue task", "what does a 5.0 essay need that a 4.0 doesn't"), wants a model essay, asks about the 30-minute timing, or wants targeted practice on position-taking, examples, or counterargument handling. Covers all six official instruction variants (agree/disagree, recommendation, claim-with-reason, two views, consequences). For IELTS essays use ielts-writing-task2; for auditing whether an argument's logic actually holds use argument-audit; for SOP/application essays use apply-sop-writer.4---56# GRE Issue Essay Coach (Analytical Writing)78The Issue task gives 30 minutes to take a position on a general-interest claim and defend it with reasons and examples. Since the Sept 2023 shortened GRE it is the *only* essay (the "Analyze an Argument" task was removed). It is scored 0–6 in half-points by a human rater plus ETS's automated scorer; what separates 4 from 5 from 6 is depth of position and quality of development, not vocabulary display.910## Use this when1112- The user shares an Issue prompt and wants a plan, a model essay, or a timed-practice setup.13- The user shares their essay and wants a score estimate with rubric-anchored feedback.14- The user asks about structure, examples, counterarguments, or the 30-minute clock.1516## Do not use this when1718- The essay is IELTS — use `ielts-writing-task2` (different rubric, different task logic).19- The user wants their own real argument checked for validity — use `argument-audit`.20- The essay is a statement of purpose or application essay — use `apply-sop-writer`.2122---2324## Task facts2526- 30 minutes, one prompt, typed in a plain editor (no spellcheck). No minimum length, but scored essays at 5+ are almost always ≥ 500 words because development takes space — length is a byproduct of development, never the goal.27- Every possible prompt is in the published ETS Issue pool (searchable on ets.org) — practice from the real pool.28- Each prompt carries one of six standard instruction stems. **The stem changes what a complete response is** — ignoring it caps the score:2930| Stem asks for | A complete response must |31|---------------|--------------------------|32| Agree/disagree with a statement | Take a position AND consider ways the statement might or might not hold |33| Agree/disagree with a recommendation | Discuss specific circumstances where adopting it would or wouldn't be advantageous |34| Which of two views aligns with yours | Choose one AND address both views |35| Claim + underlying reason | Evaluate the claim AND the reason offered for it |36| Consequences of adopting a policy | Argue from predicted outcomes |37| Statement with a qualification | Discuss when it holds and when it doesn't |3839> Score anchors: [`references/awa-scoring-rubric.md`](references/awa-scoring-rubric.md) — the 0–6 levels in compact form with the official-source pointer. Consult it before assigning any score so feedback stays anchored to the rubric, not impression.4041---4243## What actually moves the score4445The 4→5→6 gradient, in order of leverage:46471. **A position with texture.** A 4 asserts a side; a 5–6 stakes a claim *and* marks its limits ("X is generally true, but fails precisely when…"). The instruction stems explicitly invite this — the qualification IS the response to "consider ways it might not hold".482. **Developed reasons, few in number.** Two or three reasons argued deeply beat five asserted. Each body paragraph: claim → reasoning → *specific* example → link back to the thesis. "Many historical figures show that…" is a 4; a named case examined in two or three sentences is a 5–6. Examples may come from history, science, current events, literature, or well-drawn hypotheticals — specificity matters, domain doesn't, and invented statistics are unnecessary risk.493. **An honest counterargument.** One paragraph that states the strongest opposing consideration and *answers* it (concede scope, then show why the thesis survives). Straw men read as a 4.504. **Organization that carries the logic.** Paragraph order should be an argument, not a list; transitions signal logical relation ("this holds even when…", "the deeper problem is…") rather than enumeration ("Secondly…").515. **Language: varied, precise, and controlled.** Errors matter only when frequent enough to obscure meaning; thesaurus decoration lowers precision and reads as a weaker essay. Fix clarity before fanciness.5253## The 30-minute protocol5455| Minutes | Do |56|---------|-----|57| 0–4 | Read the stem twice. Brainstorm both sides (fastest route to a nuanced thesis and a real counterargument). Pick position, jot 2–3 reasons + one example each + the counterargument. |58| 4–26 | Write: intro (context sentence + thesis with its qualification) → 2–3 reason paragraphs → counterargument paragraph → short conclusion that restates the qualified thesis. Intro and conclusion stay short; the middle is where points live. |59| 26–30 | Proofread for typos, dropped words, sentence boundaries. No restructuring. |6061If time collapses: a complete essay with a conclusion outscores a longer torso that stops mid-thought — always land the ending.6263---6465## Workflow6667### Scoring a user's essay68691. Identify the prompt's instruction stem and check the essay answered *that* task — stem-mismatch is the first thing to flag, before any prose feedback.702. Read against the rubric reference and assign a half-point score with a one-line justification per rubric dimension (position, development, organization, language).713. Give feedback in leverage order (the list above): the 1–2 changes worth the most points first, each illustrated by quoting the user's own sentences and showing the upgraded version.724. End with:7374```75Estimated score: X.X / 676Biggest lever: <one sentence>77Next drill: <one concrete practice instruction>78```7980Calibration honesty: text-only scoring can't see the timed conditions; say the estimate assumes the essay was written in 30 minutes without tools, and ask if it wasn't.8182### Producing a model essay or plan8384- Under timed-realistic constraints: ~500–650 words, achievable structure, no encyclopedic examples the user couldn't produce live.85- Label the parts (thesis, reasons, counterargument) and annotate *why* each paragraph earns its keep, so the user learns the skeleton, not the specific essay.8687### Timed practice8889Offer prompts from Issue-pool themes (society, education, technology, government, intellectual life), enforce 30 minutes, then score as above. Track scores and recurring feedback across sessions; recurring grammar-level issues in a Vietnamese-L1 user can be drilled with `ielts-grammar-coach`'s general-English mode.
jurgendn