# Sources, attribution, and distribution

## Word inventory
ENABLE (Enhanced North American Benchmark Lexicon), associated with Alan Beale and Mendel Cooper. Source snapshot: https://raw.githubusercontent.com/dolph/dictionary/master/enable1.txt retrieved 2026-10-13 UTC. ENABLE's public-domain status is documented in the included ENABLE-LICENSE-SOURCE.txt, from https://github.com/en-wl/wordlist/blob/v2/Copyright.

## Familiarity
wordfreq 3.1.1, by Robyn Speer and contributors: https://github.com/rspeer/wordfreq and https://pypi.org/project/wordfreq/3.1.1/ . ENABLE words of 3–8 letters were given wordfreq English Zipf scores and sorted into familiarity tiers. Frequency is a rough familiarity proxy, not proof that a word suits all ages.

The upstream wordfreq program license is Apache-2.0 (see WORDFREQ-LICENSE.txt). Its data includes material redistributed under CC BY-SA 4.0; full attribution is preserved in WORDFREQ-NOTICE.md. Credit Robyn Speer and contributors; Marc Brysbaert et al. for SUBTLEX, freely available at http://crr.ugent.be/programs-data/subtitle-frequencies ; OpenSubtitles at https://www.opensubtitles.org/ ; Google Books Ngrams at https://books.google.com/ngrams ; and the other sources named in the full notice.

## Definitions and categories
Princeton WordNet 3.0, as packaged in the Open Multilingual Wordnet English release (omw-en 1.4) and read with the `wn` Python package. WordNet 3.0 Copyright 2006 by Princeton University. All rights reserved. Used under the WordNet 3.0 licence, whose full text, including its disclaimer, is in WORDNET-LICENSE.txt. WordNet supplied definitions, example sentences, synonyms, lexical categories and meaning frequencies. Clues were shortened, rewritten or replaced by hand from those glosses; every puzzle answer in the game uses a hand-written clue.

## Starter clues
595 friendly full-sentence clues for common words come from the reviewed 612-word starter set in `data-sources/starter-612/` (written for this project; see its README and SOURCES). They replace the shorter clues for those words (`tools/import_starter_clues.py`).

## This derived data
The word bank (`wordbank/`), dictionary.json, levels.json, campaign.json and answers.csv are distributed under CC BY-SA 4.0: https://creativecommons.org/licenses/by-sa/4.0/ (full text in CC-BY-SA-4.0.txt). Retain this attribution, identify your modifications, link or include the licence, apply its ShareAlike terms to adaptations of these data, and keep the WordNet notice with them. This notice doesn't require the separate game code to use the same licence.

Modifications made for this game: words limited to 3–8 ASCII letters; familiarity tiers; exclusion of names, slurs, explicit terms and their forms; mature-theme flags; short clues written from WordNet glosses (thousands rewritten by hand); categories and topics mapped from WordNet's lexicographer files; generation of 1,000 boards from 3×3 to 8×8 with unique letter sets, a word-reuse gap and starter letters; validation.

Suggested in-game credit: "Word list: ENABLE (public domain). Familiarity: wordfreq by Robyn Speer and contributors, including SUBTLEX by Marc Brysbaert et al. and OpenSubtitles. Definitions: Princeton WordNet 3.0, © 2006 Princeton University. Puzzles and clues: CC BY-SA 4.0." Link to this file and the notices.