kidplayclub.com Word Tiles dictionary — modified SCOWL distribution This is an original game dictionary selection, not an official Scrabble, NASPA, Collins or other proprietary tournament word list. Those publishers do not endorse it. Game rules must use the dictionary ID saved with a match; changing any membership policy requires a new dictionary ID. Source and license ------------------ Source: Spell Checking Oriented Word Lists (SCOWL), Kevin Atkinson and contributors, release 2020.12.07, https://wordlist.aspell.net/ . Source repository: https://github.com/en-wl/wordlist . Pinned release commit: 5ef55f9c42730ebe4394a78b77855468a6f15dd2 . The release archive, SHA-256, individual input hashes, output hash and precise transformation are recorded in word-tiles-provenance.txt. This modified distribution retains the complete, verbatim upstream Copyright file as word-tiles-scowl.txt, including all constituent notices. The collective license is permissive, MIT-like (not simply SPDX MIT); constituents include public-domain material, the Princeton WordNet license, the UKACD attribution license, and VarCon/Ispell permissive terms. Do not remove these notices when redistributing the generated data, including binary/browser distributions. Make these files available with the product documentation. Individual words are not claimed as exclusive kidplayclub.com property. Deployed license documents (exact mirrors of licenses/word-tiles*.txt): /licenses/word-tiles-policy.txt /licenses/word-tiles-scowl.txt /licenses/word-tiles-provenance.txt /licenses/word-tiles-exclusions.txt The exclusions file contains offensive source terms for adult audit only; link the policy and license, rather than presenting exclusions to children. Prominent UKACD notice: Copyright (c) J Ross Beresford 1993-1999. All Rights Reserved. Special source credits: Kevin Atkinson; Alan Beale (12Dicts and SCOWL supplement); M. Cooper and contributors (ENABLE); Grady Ward (Moby); Brian Kelk; J Ross Beresford (UKACD); Princeton University (WordNet); Benjamin Titze and Geoff Kuenning. Credit is not endorsement. Selection and limits -------------------- 236,728 distinct uppercase ASCII words, 2 through 15 letters inclusive. Only SCOWL's normal "words" category, English + American spellings, levels 10, 20, 35, 40, 50, 55, 60, 70 and 80. SCOWL documents level 80 as including unusual words used in word games. This gives broad playable coverage, but includes rare, dated and specialist words; it is not a spelling curriculum. It is US-oriented, not a promise to include every British spelling. Only original lowercase ASCII a-z entries are accepted before uppercasing. No names/upper/abbreviations/contractions/special/variant categories are imported. No punctuation, spaces, accents, accent folding or invented inflections are added. A lowercase common-word homograph of a name may still be valid. A valid English word absent from this selection is rejected. One-letter words and words longer than a 15-cell board are not permitted. Automatic policy: every newly formed main word and perpendicular cross word must be in this exact versioned dictionary. Invalid words are rejected automatically without a human challenge or generative AI. Lookup accepts exact uppercase only; it does not silently trim or change case. Child-content limitation — not educator reviewed ---------------------------------------------- This broad list has NOT been reviewed or approved by educators, and is NOT a certified child-safe dictionary. Adult supervision is recommended. The generator excludes exact normalized entries in the upstream files misc/profane.1, misc/profane.3, misc/offensive.1 and misc/offensive.2. Those four files supply 81 distinct A-Z entries of 2..15 letters; 77 occurred in the selected dictionary and were removed. Their normalized forms are preserved in word-tiles-exclusions.txt for reproducibility, not for presentation to children. SCOWL describes profane.1 as vulgar/swear words inappropriate around children; profane.3 as less strong taboo/vulgar words; offensive.1 and offensive.2 as racial slurs of differing severity. These are small, historical lists, not comprehensive moderation. Exact-match removal can remove innocuous homographs while missing inflections, compounds, other slurs, adult topics or context-sensitive meanings. There is no substring filter and no claim of complete offensive-word exclusion. WORD_TILES_BOT_WORDS is an independently selected, bounded ordinary-word subset checked against the same dictionary. It is not generated by an AI service during play and cannot change which human words are valid. Reproduction and verification ---------------------------- Run offline checks (no dependency installation or network required): node --experimental-strip-types scripts/validate-word-tiles-lexicon.mts To verify source membership as well, download the pinned archive URL in word-tiles-provenance.txt to a project-local path (never a system temporary directory), then run: node --experimental-strip-types scripts/validate-word-tiles-lexicon.mts PATH_TO_ARCHIVE The optional source check verifies the archive and each selected input hash before reproducing the exact membership transformation. It never executes upstream scripts or extracts archive paths to disk. The generated word module is a frozen newline-split string containing that sorted result; hashing the result joined with LF plus one terminal LF must match wordsSha256. No network access occurs in the runtime dictionary.