Verified against ChatGPT · 2026-08-09
Build a grammar drill set targeted at the one mistake you keep actually making
Generates a focused set of practice sentences for one specific grammar point you're struggling with in a language you're learning, sized to your real level instead of a generic textbook chapter's worth of exercises.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Build me a targeted practice set for one specific grammar point in Spanish that I keep getting wrong, not a general review of the whole topic area. MY LEVEL Second-year high school Spanish, comfortable with present tense, shaky on past tenses. THE SPECIFIC GRAMMAR POINT Choosing between preterite and imperfect past tense. EXAMPLES OF MY ACTUAL MISTAKES I wrote 'Cuando era niño, fui a la playa todos los veranos' when it should have been 'iba' — I keep using preterite for repeated past actions. HOW MANY ITEMS AND IN WHAT MIX 12 items total: 5 fill-in-the-blank, 4 sentence correction, 3 translation. Build the set in three parts. First, a one-paragraph explanation of the rule stated in terms of the specific mistake pattern I showed you — not the textbook definition, but why my exact error happens and what distinguishes the correct form from what I keep writing. Second, the practice items themselves: a mix of fill-in-the-blank, sentence correction (where you give me a sentence with the mistake already in it and I have to fix it), and translation-direction items, matching the mix I specified. Make each item test the actual distinction I'm confusing, not a broader area of the grammar point I already have right — if my mistakes show I've confused two specific forms, every item should force a choice between exactly those two, not drift into testing the grammar point's other, unrelated edge cases. Third, an answer key with a one-line reason per answer that references the same distinction from your explanation, so a wrong answer sends me back to the same explanation rather than a new one. If my example mistakes actually show two different underlying confusions rather than one, tell me that directly and ask which one to build the set around, rather than building one set that tries to cover both and ends up testing neither well. Deliver the set as: (1) the explanation paragraph, (2) numbered practice items grouped by type, (3) the answer key.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
A prompt that just says "give me practice with preterite vs imperfect" gets GPT-5.1 to generate a broad, textbook-shaped set that samples across the whole grammar topic evenly, because without a stated error pattern the model has no signal for which sub-distinction actually matters to you and defaults to comprehensive coverage as the safest-seeming answer. Feeding it your actual wrong sentences changes the task from "teach this grammar point" to "resolve this specific confusion," which is a narrower and more useful target — the model can identify that your error is specifically about habitual/repeated past actions being marked with preterite instead of imperfect, and then construct every item to force exactly that choice, which drills the actual gap far faster than a set that also tests preterite-vs-imperfect distinctions you already have right. Asking the model to check whether your mistakes reveal one confusion or two matters because self-diagnosed grammar mistakes often bundle multiple issues that look similar on the surface (e.g., aspect confusion versus irregular-verb-conjugation errors) but need different drill types to fix; a model that silently picks one interpretation and builds a sanitized single-purpose set may drill the wrong thing entirely, while surfacing the ambiguity lets you correct it before generating twelve items that don't address your real problem. Tying each answer-key explanation back to the same opening rule, rather than a fresh ad hoc justification per item, keeps the whole set coherent as one lesson instead of twelve disconnected trivia facts, which is what actually gets a distinction to stick after repeated exposure.
What you get back
3. Fill in the blank: De niña, mi abuela ______ (ir) a la iglesia todos los domingos. Answer: iba — repeated/habitual past action ('every Sunday') takes imperfect, not preterite, matching the same pattern from your 'fui a la playa todos los veranos' mistake.
Verified against
ChatGPT GPT-5.1 · 2026-08-09
Changelog
- 2026-08-09 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
