Mundus patet, verbum patet

The world lies open, the word lies open — notes towards a Semantic Archaeology. (Repository name: dormant.)

A curated lexicon of forgotten images. Each entry replaces a common English word with the literal meaning of its etymological roots (window → wind eye, mortgage → death pledge), making ordinary prose briefly strange without becoming unintelligible.

297 verified entries as of 2026-10-02 (plus 19 derived forms such as government, president and company), spanning Old English/Norse household and kinship terms, Latin/Greek abstractions, law & religion, medicine, trade & textiles, and measurement. Two high-frequency passes — one on 19th-century novels (world → age of man, answer → counter-oath), one on modern English (video → I-see, worry → strangle, phone → far voice) — raised coverage of modern prose to about 40 recovered words per 1,000. See research_notes.md for the session log and open research domains.

The piece

The piece has two modes, side by side under the title: Read and Write. The title is from the Roman mundus patet, "the world is open", the three days a year the pit of the dead was uncovered; a footer note on both pages explains it.

Read (docs/index.html). Open it in a browser: paste your own prose or pick a sample passage, and it renders a translation (roots recovered) above your original text. Hover a recovered word to see the source word and its etymon. Contested words are marked but never glossed. In the default Stir view the text starts as plain English and changes in place as the mouse passes over it — each word sinks to its root and surfaces as its image (window → vindauga → wind eye), so the reader stirs a new poem out of the prose. Three sample passages are built in: The household, The scholar and The sentence.

Write (docs/write.html, prototype). The writer composes a poem in which every line holds a lexicon word: from a dealt hand, or from the whole lexicon, which is listed by image family with a find box. When every line holds, the poem bleeds (images rise as ghosts over the words) or stirs.

Research continues in three directions: homonyms (one spelling, unrelated words: homonyms/), root families (entries that share a root wake together; the optional root field), and hedged theories for contested words (sourced, always marked "perhaps"). See Start Here at the end of research_notes.md.

Runs from file:// with no server — just open docs/index.html. The same folder is what GitHub Pages publishes (Settings → Pages → main / docs), so the piece is also live on the web; nothing else in the repository is served.

Files

• docs/index.html — Read mode, the electronic-literature piece, and what Pages serves
• docs/write.html — Write mode, constraint writing (prototype): a poem in which every line must hold a word from a dealt hand or the whole lexicon; finished, it bleeds (images rise as ghosts over the words) or stirs
• docs/stir.js — matching and stirring engine shared by both pages
• docs/lexicon.js — generated lexicon data (run python3 build_lexicon.py from the repository root to rebuild from canonical.json)
• build_lexicon.py — validates canonical.json (required fields, score ranges, contested entries unglossed, unique words, sources on every entry and form), then bakes it into docs/lexicon.js; --check validates only
• canonical.json — accepted, verified entries (production lexicon)
• candidates.json — research queue (verified candidates carry scores and a verdict in editorial_notes)
• editorial.md — editorial policy, criteria, and methodology
• poetics_of_semantic_archaeology.md — conceptual essay
• research_notes.md — laboratory notebook
• homonyms/ — homonym collection: Wiktionary finder and filter scripts, the 374-word shortlist, the handoff notes