French word list - origin and licence
=====================================

Source:    Lexique des formes fléchies du français, Grammalecte v7.7
           Olivier R. / Dicollecte - https://grammalecte.net/
           https://www.grammalecte.net/dic/lexique-grammalecte-fr-v7.7.zip
           SHA-256 of the archive: bdccf2065e252010253171000a38de84581407896b6bc60aca7d4d69dab7e4d2
Retrieved: 2026-09-27
Licence:   Mozilla Public License 2.0 (MPL-2.0), full text in MPL-2.0.txt.
           Original notice of the lexicon: README_lexique-grammalecte.txt.
           The unmodified source file is available at the URL above.

The word lists in this folder are an extract (a modified form) of that lexicon,
made with scripts/wordlist-fr-extrait.py (Python 3, PYTHONUTF8=1):

  python scripts/wordlist-fr-extrait.py lexique-grammalecte-fr-v7.7.txt
    -> vendor/wordlists/fr/lexique-fr-extrait.txt ("word<TAB>frequency index")

Filters:
  - only lower-case words a-z with French accents and ligatures (à â ä ç é è ê ë
    î ï ô ö ù û ü ÿ œ æ), 2-20 characters: removes proper nouns, hyphenated
    words, forms with apostrophes, abbreviations with dots
  - without entries tagged as proper nouns (npr, prn)
  - without entries noted as pejorative ("péj", includes slurs), abbreviations
    ("abr"), acronyms ("sig") and symbols ("symb")
  - sub-dictionaries "*" (common), "M" (modern) and "R" (1990 spelling reform);
    classical-only ("C") and other variants are left out
  - frequency index >= 1 (index 0 = form not attested in any of the corpora
    Grammalecte uses)
  - all inflected forms (plurals, feminine forms, conjugated verbs) are kept

Then scripts/build-wordlist.mjs writes public/data/words/fr/<length>.txt (only forms with
frequency index >= 3: 215,174 forms, each file at most about 360 KB) and wordle.txt. The word finder and the Wordle helper ignore accents and treat
ligatures as two letters (É = E, Ç = C, Œ = OE), as in Scrabble, crosswords
and French Wordle games. wordle.txt contains the words with exactly five
letters after that conversion, a frequency index >= 4 (of 9) and Zipf >= 3 in the
wordfreq frequency list (CC BY-SA 4.0, see README_wordfreq.txt), so that only common words
are recommended as start words: 2,285 entries, 2,090 distinct words. Because of the wordfreq
data, wordle.txt is distributed under the GNU GPL version 3 or later (COPYING-GPL-3.0.txt);
the <length>.txt files remain under MPL-2.0.