denckring-en-data
Copyright 2026 senzelden

The code in this distribution is licensed under the Apache License, Version 2.0.
See LICENSE.

The data in this distribution is not, and is redistributed under its own terms:

  - src/denckring_en_data/data/cmudict.dict is the CMU Pronouncing Dictionary,
    Copyright (C) 1993-2015 Carnegie Mellon University, under the BSD-2-Clause
    terms reproduced in LICENSE-CMUDICT.

  - src/denckring_en_data/data/nouns.txt and data/glosses.txt.gz are derived from
    the Open English WordNet 2024 (oewn:2024), licensed CC BY 4.0 and reproduced in
    LICENSE-WORDNET. Both files are modifications: nouns.txt is a filtered list of
    single-word noun lemmas and glosses.txt.gz a compressed lemma-to-definitions
    index, neither in OEWN's own format. scripts/build_lexicon.py is the derivation.

    Open English WordNet, https://en-word.net/, CC BY 4.0

  - src/denckring_en_data/data/graded_words.txt.gz is derived from SCOWL
    2020.12.07, the collective work Copyright 2000-2018 Kevin Atkinson, under the
    terms reproduced in LICENSE-SCOWL. That file is SCOWL's whole Copyright file
    rather than an excerpt of the parts this build draws on, because Atkinson's
    permission notice asks that both the copyright notice and the permission
    notice appear in supporting documentation, and because a licence excerpted to
    the sources drawn on today would need re-excerpting the next time the size
    cap moves. UKACD, whose terms require its notice be reproduced verbatim
    wherever it is redistributed, enters SCOWL at the 80 level and so is not
    redistributed here at all.
    graded_words.txt.gz is a modification: SCOWL's `english-words` and
    `american-words` lists at sizes 10 through 60, reduced to one `word<TAB>size`
    line per word and in none of SCOWL's own formats.
    scripts/build_graded_words.py is the derivation.

    SCOWL, http://wordlist.aspell.net/, see LICENSE-SCOWL

  - src/denckring_en_data/data/frequencies.txt.gz is derived from the Leipzig
    Corpora Collection's eng_news_2023_1M corpus, licensed CC BY 4.0, with the
    attribution in LICENSE-LEIPZIG. It is a modification: the corpus's word list
    reduced to the lowercase forms of the words graded_words.txt.gz holds, one
    `word<TAB>count` line each. No sentence from the corpus is included.
    scripts/build_frequencies.py is the derivation. The licence version is
    established as for denckring-de-frequency (ADR 0038 D1): the Leipzig terms say
    "CC BY", and the link on those words points at the 4.0 deed.

  - src/denckring_en_data/data/everyday_exclusions.txt is this project's own
    editorial list (ADR 0051), under the Apache License like the code.
