Retrofit the spoken-punctuation pamphlet

[dcellison]

Summary

• Rebuild the spoken-punctuation pamphlet around Phi's actual grammatical boundaries, treating punctuation language as a teaching analogy rather than a second grammatical system.
• Correct the lessons on direct and embedded questions, quotation, names, clause boundaries, emphasis, and dictation, with exercises and answer keys that now follow current canon.
• Update the pamphlet catalogue, SEM-09D tracking, the thilou usage note, and the phonetic-neighbour baseline to match the revised teaching corpus.

What changed

The question lesson now distinguishes initial wa in direct yes-or-no questions from the seven content-question gap words and from embedded wela ... welo clauses. It also replaces an improvised wonder construction with the established phaelo pattern.

The quotation lesson now keeps exact Phi wording inside shola ... sholo, uses the appropriate speaking, asking, and hearing frame verbs, and treats material in another language as source material outside the Phi sentence. The name lesson explains portable and local ne usage without making physical presence a condition.

The clause lesson identifies construction-specific boundaries and includes an explicit modifier-first purpose-clause example. The emphasis lesson explains that Phi has no general equivalent of an exclamation mark: ru, su, teo, interjections, and prosody retain their own meanings instead of becoming interchangeable punctuation words.

The dictation lesson now separates recoverable lexical and grammatical information from silent written period placement. It also corrects the motion endpoint to kau lo mia and uses the base word sukaro for intense heat instead of the earlier workaround.

The exercise set and appendix have been rebuilt around those distinctions. No vocabulary roots, module assignments, or lexical meanings were added. The only lexicon edit corrects thilou's usage note, which had incorrectly described direct questions as using final wa.

Project records now mark three of nine SEM-09D pamphlets complete and identify pamphlets/complementizers/ as the next retrofit. The phonetic-neighbour baseline was regenerated because the revised examples changed corpus attestation counts, not because the lexicon changed.

Voice review

The Humanizer and Phi voice pass removed inflated claims that Phi's punctuation system made speech calm or prevented shouting. It also replaced repeated matrix-predicate phrasing, compressed inventory-heavy prose into readable tables and lists, and removed speed-oriented drills while preserving every validated Phi form.

Checks

• python3 scripts/validate_examples.py --show-warnings (1,275 entries, 0 errors, 0 warnings)
• python3 scripts/test_vocabulary_schema.py (22 tests)
• python3 scripts/content_vocabulary_decisions.py --check (63 batches, 235 decisions)
• python3 scripts/test_content_vocabulary_decisions.py (6 tests)
• python3 scripts/test_name_forms.py (20 tests)
• jq empty pamphlets/catalogue.json vocabulary/content/inquire.json
• python3 scripts/generate_reference.py (no generated-reference diff)
• python3 scripts/audit_phonetic_neighbors.py --output /tmp/spoken_punctuation_phonetic_neighbors_final.txt (250 pairs; exact match with the updated baseline)
• python3 scripts/build_site.py (complete site build, including 98 pamphlet pages)
• git diff --cached --check