The generator validates one to twelve supplied records, lays out word-and-ID fronts in six-slot groups, and reveals a separate teacher sheet. Reversing the order keeps the notes with their IDs. The sample words said and the stay labeled Draft: teacher review pending; this is an HTML proof rather than a completed DOCX or verified print artifact.
Put instruction before the practice proof
UFLI distinguishes correspondences not yet taught from unexpected spellings, and places flashcard practice after instruction in the relevant parts. Use that instructional distinction without copying its deck, artwork, heart marks or exposure-count/learning-gain claims. UFLI: Irregular and High Frequency Words
The input should be a teacher-supplied record, not just a word list. For each stable ID, retain word, spoken-target reference, supplied spelling chunks, teacher notes, per-chunk Taught/Not yet taught/Unexpected labels, and guidance source/date/confirmation.
Those labels describe current instructional context. Do not infer them from a universal irregular-word list. Word frequency, a supplied chunk list and an attractive front do not establish what this class has learned.
Spoken-target reference is a source or description supplied by the teacher. The tool does not synthesize audio, generate phoneme/IPA transcription or score pronunciation. Leave absent references TO CONFIRM.
Validate spelling without approving instruction
| Fictional record | Word | Supplied chunks | Structural check | Guidance status |
|---|---|---|---|---|
| W01 | said | s / ai / d | Concatenation equals said | Real teacher confirmation pending |
| W02 | the | th / e | Concatenation equals the | Context labels/notes unconfirmed |
The lists are illustrative fields, not approved phoneme mappings. s+ai+d reconstructs said; th+e reconstructs the. This mechanical check says nothing about intended spoken form, sound value or which parts are taught.
The generator checks that IDs are present and unique and that the supplied chunks reconstruct the exact spelling and case of each word. W01 with s / a / d gives sad, not said; reject that structural record rather than silently inserting the missing letter.
Accept one through twelve records. Words have at most twenty ASCII letters/apostrophes and must contain a letter; notes have at most 120 characters per field. These are local bounds, not a vocabulary or curriculum standard.
Unconfirmed guidance differs from a malformed record. Structurally valid W02 can appear in a draft teacher proof with missing fields TO CONFIRM. It is not practice-ready. Do not automatically complete labels, source/date or approval.
Match fronts and guidance by ID
The student front contains the supplied word and record ID. A separate teacher sheet carries that same ID, word, spoken-target reference, chunks, labels, notes, source/date and confirmation state.
Changing order cannot move W01's notes onto W02. Do not match by position, rely on row number alone or assume duplex alignment. This packet deliberately uses a separate teacher sheet rather than alignment-dependent backs.
The learner's task is to read the word after relevant instruction, not answer a definition question on a back. There is no automatic recognition score, learned flag or mastery indicator. Real teacher observations belong in the teacher's own record.
Both examples remain Draft: teacher review pending. A valid spelling structure does not confirm the instructional guidance.
Generate and inspect the proof
The editable JSON sample contains W01, said, with chunks s / ai / d, and W02, the, with th / e. Press Generate proof to see the two fronts. Four empty slots remain in their six-slot group; they are deliberate blanks, not missing records. Reveal teacher sheet to compare ID, word, chunks, notes and source status.
Change W01’s chunks to s / a / d and generate again. The structure is rejected because sad does not reconstruct said. The old proof is cleared, so a stale card cannot appear validated. Restore the sample before reviewing the layout. Reverse input order, regenerate, and verify that W02 keeps its own notes.
Spelling reconstruction is a mechanical check. Keep spoken-target references, class-specific labels and teacher confirmation in your maintained teaching file. The local proof does not create audio, phoneme transcriptions or recognition scores. A six-slot screen arrangement still needs document and physical-print review before classroom cutting or use.
Copy this prompt
Build an HTML word-front proof generator for one to twelve teacher-supplied JSON records. Retain supplied spoken-target references, per-chunk Taught/Not yet taught/Unexpected labels, guidance date and confirmation text; missing guidance remains TO CONFIRM and the proof stays pending review. Each record has stable unique ID, word, an array of nonempty spelling chunks, teacher notes and source reference. Sample W01 word said, chunks s/ai/d, notes “Illustrative spelling chunks; real teacher guidance pending”, source TO CONFIRM. W02 word the, chunks th/e, notes “Instructional labels and spoken target TO CONFIRM”, source TO CONFIRM. These are fictional unconfirmed records, not approved phoneme mappings. Optional spokenTarget, labels, guidanceDate and confirmation fields carry teacher-supplied text. Labels must provide one Taught, Not yet taught, Unexpected or TO CONFIRM value per chunk. Show absent fields as TO CONFIRM and keep the overall draft label even when guidance is supplied.
Validate unique nonempty IDs, ASCII-letter/apostrophe words of at most twenty characters containing a letter, nonempty chunks and exact case-preserving chunk concatenation. Reject W01 chunks s/a/d because they make sad. Bound note and source fields to 120 characters. Generate proof builds front cards containing only word and ID, grouped into six slots, two columns by three rows. Mark unused slots Blank slot. A separate Reveal teacher sheet displays matching ID, word, chunks and supplied guidance. Twelve records require two planned groups; two sample records use one.
Provide editable JSON, Generate proof, Reverse input order, Reveal teacher sheet and Clear. Editing invalidates old proof. Reordering preserves ID-to-guidance association. Keep Draft—teacher review pending visible; neither structure nor notes reveal can grant approval. This is a browser proof, not a downloadable PDF/DOCX or verified printed pagination.
Return one complete HTML file. Show the JSON editor before the word-front preview and keep the teacher sheet separate. Use readable words, visible record IDs and six slots per group, leaving unused slots visibly blank. On a phone, stack the editor and preview without creating extra groups. Only supplied guidance should appear on the teacher sheet.
Review guidance before preparing classroom copies
Keep the editable front packet and teacher guidance sheet in your teaching files. Confirm the spoken targets and current instructional labels before preparing practice materials, then check the actual print size and cut spacing. For the teacher’s guidance review, NOS can render the prepared HTML fronts beside their ID-linked notes. Review W01 and W02 by ID after reordering so a neat layout never hides a mismatched note.