Reproducible Word Data Studies
Original analyses generated from the same pinned lexical release that powers the product. A checksum-valid catalog can be prepared without publishing its results; numbers and downloads appear only after the required human reviews approve the exact artifact.
The study catalog is validated; all outputs are withheld. Required factual, editorial, and source-rights approvals are incomplete, so no result, metadata bundle, or download is public.
Published studies
These source-bounded questions have a published lifecycle record. That record does not expose a result: the exact artifact remains unavailable until all required review dimensions approve it.
How three English spelling lists overlap
The comparison shows why presence in one list is evidence of source membership, not universal English validity.
Awaiting required human reviewWhat supports The Word Index commonness tiers
Users can judge what a commonness label means and avoid treating an estimated tier as a universal fact.
Awaiting required human reviewPronunciation coverage in the compiled lexicon
Coverage figures show which retained records have CMUdict evidence for rhyme and syllable tools; absence of that evidence does not imply that a spelling lacks a pronunciation.
Awaiting required human reviewLetter positions by spelling length and frequency tier
The study distinguishes patterns in the broad five-tier cohort from patterns in its higher-ranked tier 1-2 cohort.
Awaiting required human reviewAnagram-family density in broad and higher-ranked vocabularies
The comparison shows how apparent anagram choice differs between the broad five-tier cohort and its tier 1-2 subset; neither cohort establishes universal or game-list validity.
Awaiting required human reviewHow publication works
A finite version-controlled registry defines the question, sources, criteria, calculations, limitations, review state, and reproducibility command. The build creates canonical JSON and CSV once. The web app verifies their checksums and exact lexical-release binding; it never recalculates a missing result.
Current study release: 57cd6e4dac25b8a1.
Review the lexical source and licence register and the wider data methodology.