# Source attribution and data terms ## Open English WordNet 2025 The local meaning-association index uses [Open English WordNet](https://en-word.net/), copyright the Open English WordNet Community, derived from Princeton WordNet, under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). The pinned 2025 XML release is downloaded from https://en-word.net/downloads/english-wordnet-2025.xml.gz. Source and compiled checksums are recorded in `data/wordnet/manifest.json`. Modifications: retain lowercase lexical forms, definitions, synonym sets and direct hypernym relations; omit examples and senses explicitly marked offensive; compile a compact local lookup. This collection supports matching only and does not activate naming words or infer translations from local-language spelling. Attribution is also served at `/wordnet-credits.txt`. **English cue words, 10 November 2026:** 100 dictionary entries in `data/imports/sa-cues-review.json` take their meaning word for word from Open English WordNet 2025 definitions. Each entry's source field names the synset and CC BY 4.0. Modifications: selected one sense per word and added provisional naming tags; no definitions were written or paraphrased. See [the cue word report](SA-CUES-2026-11-10.md). **South African flavour collection, 10 November 2026:** 43 dictionary entries in `data/imports/sa-flavour-review.json` take their meaning word for word from Open English WordNet 2025 definitions of southern African plants and animals. Each entry's source field names the synset and CC BY 4.0. This is the one place where WordNet text becomes a naming-entry meaning; those entries are Needs review until an administrator activates them. A further 93 entries in the same batch use Wiktionary senses through Kaikki (South African English regional senses and Afrikaans) under CC BY-SA 4.0, each with its page link; contributor-history links are kept in `sa-flavour-provenance.json`. Modifications: selected one sense per word and added provisional naming roles and tags; no definitions were written or paraphrased. See [the flavour collection report](SA-FLAVOUR-2026-11-10.md). This file accompanies the local source collection, prepared datasets and imported review entries. Preserve it, the source registry, checksum lock and original notices when copying the data. The sources have different terms; they are not a single public-domain word bank. ## Four-language Wiktionary noun snapshot and second expansion **Wiktionary contributors**, English Wiktionary Tswana, Tsonga, Venda and Northern Sotho noun entries, snapshot 10 November 2026. Source category membership and page revisions are retained under `raw/wiktionary-sa-nouns-2026-11-10/`; derived senses retain page, fixed revision and contributor-history URLs. Original dictionary text is **CC BY-SA 4.0**, subject to the [Wiktionary copyright policy](https://en.wiktionary.org/wiki/Wiktionary:Copyrights) and [licence](https://creativecommons.org/licenses/by-sa/4.0/). Quotations and media are excluded from the derived text. The derived `prepared/wiktionary-nouns/`, `data/imports/sa-round2-review.json`, `sa-round2-provenance.json`, and their entries in MySQL are CC BY-SA 4.0. The second batch also selects previously unused equivalents from the Kaikki, UP and IsiXhosa.click snapshots credited below. Modifications: selected complete English equivalents, unwrapped recognised wiki formatting, retained original spelling, raw definitions, labels, grammar and revision evidence, and added provisional naming tags and review status. No translations or missing grammar were generated. Preserve source-specific credit, history/revision links, licence and modification notices; distribute adapted dictionary data under the same licence. ## African Wordnet Griesel, Marissa. **African Wordnet version 1.0**. UNISA, 2022. Distributed by SADiLaR: . Requested project citation: Bosch, Sonja and Griesel, Marissa. 2017. *Strategies for building wordnets for under-resourced languages: the case of African languages*. Literator 38(1). . The 2022 repository record and project distribution page identify **Creative Commons Attribution 4.0 International**: . The saved repository metadata, licence page and archive README provide the evidence. The nine African-language prepared files retain this attribution and licence. Modifications by the name-za project, 9 November 2026: converted XML records to JSONL, separated synsets, added source references and review status, and mapped the South African isiNdebele code to `nr` while retaining the original `nd`. No definitions or translations were supplied by this project. Source spellings and examples are preserved. The archive also bundles English WordNet. It is not used in prepared collections or imports. That content has its own Princeton WordNet notice: . Do not apply the African Wordnet licence to it by assumption. The original ZIP remains available locally as acquired. ## Afrikaans and South African English Hunspell dictionaries Original sources: LibreOffice `dictionaries` repository at commit `c0865a85eb67dc377f2cf8e8fe2c4fa9e53f0dcb`: - Afrikaans: . Credits include Bernard A Nieuwoudt, Danie Viljoen, Dwayne Bailey, Reinier de Vos, Kris van der Merwe, PuzzleFoundry and Translate.org.za contributors. - English (South Africa): . Credits include Kevin Atkinson, David Bartlett, Brian Kelk, Andrew Brown, Dwayne Bailey, Zuza Software Foundation, Marco A.G.Pinto and contributors. The original `.dic`, `.aff` and `README_*` files are retained under `raw/`. The README files include the **GNU Lesser General Public License version 2.1** and detailed contribution notices. The Afrikaans affix file explicitly specifies version 2.1 or later. Preserve the exact notices and applicable terms, including source availability when distributing covered modifications: . Modifications by name-za, 9 November 2026: represented dictionary lines as JSONL, separated flags and morphology, retained each original line, and added source references and an unreviewed status. Affixes were not expanded. The spelling datasets remain separate from the meaning-based naming dictionary and are not imported into it. ## Wiktionary through Kaikki and Wiktextract Text authored by **Wiktionary contributors**, extracted by **Kaikki.org / Wiktextract, Tatu Ylonen**. Extraction reference: Tatu Ylonen (2022), *Wiktextract: Wiktionary as Machine-Readable Structured Data*, Proceedings of LREC 2022, pp. 1317–1325. Source distributions: , , . Original page and contributor-history URLs accompany every prepared sense and every imported entry. English Wiktionary's original entry text is available under CC BY-SA 4.0 / GFDL; this project uses the **CC BY-SA 4.0** option. Copyright policy: . Licence: . Source quotations and media may have different rights; they are excluded from the prepared and imported data. Original snapshots are retained locally for source inspection. The derived `prepared/kaikki/` files, `data/imports/sa-review-starter.json`, `data/imports/sa-review-provenance.json`, and their sourced entries copied into `data/dictionary.json` are supplied under **CC BY-SA 4.0**. Retain attribution, source/history links, the licence link and modification notices; distribute adaptations of this material under the same licence. This notice concerns sourced dictionary data, not a blanket licence assignment to unrelated project code or original branding assets. Modifications by name-za, 9 November 2026: selected lexical senses, omitted examples/quotations/media/etymologies, added provenance and review status, selected a naming shortlist through complete English-gloss matches, and added provisional editorial tags from the existing starter dictionary. Source glosses were not translated or rewritten. The source dictionaries and extractors do not endorse the resulting names. This also applies to the additional Siswati, Sesotho and Southern Ndebele Wiktionary snapshots under `raw/kaikki-sa-expansion-2026-11-09/` and their derived records in `data/imports/sa-review-expansion.json` and its provenance file. Page and contributor-history links accompany the source senses. ## University of Pretoria multilingual lexicons, version 2 **Sindane, Thapelo; Marivate, Vukosi; Modupe, Abiodun. South African multilingual lexicons, version 2.** University of Pretoria research data repository. . The primary repository metadata identifies **CC BY-SA 4.0**: . The metadata and downloaded original file are preserved under `raw/up-multilingual-lexicons-v2/`. Derived `prepared/up-lexicons/` records, their review entries in `data/imports/sa-review-expansion.json` and `data/dictionary.json`, and their provenance retain that licence. Retain credit, source URLs, licence link and modification notices when copying or adapting them. Modifications by name-za, 9 November 2026: expanded each supplied alternative into a separate traceable pair; retained source keys/row/alternative indexes and original spellings; trimmed surrounding whitespace and NFC-normalised candidate spellings; selected complete English-equivalent matches to the starter vocabulary; added provisional tags and review status. No translation, missing part of speech or full definition was generated. Equivalent pairs may contain phrases, inflections and sense ambiguity. ## IsiXhosa.click database **IsiXhosa.click contributors**, including the contributors and underlying-source acknowledgements linked at . Each selected record retains its original word-page URL and notes in the provenance file. Database snapshot: . The publisher's About page explicitly licenses the database under **CC BY-SA 4.0**: . The source CSV, README and About page are retained under `raw/isixhosa-click-2026-09-26/`. Derived `prepared/isixhosa-click/` records, expansion review entries, provenance and copies in the live dictionary retain CC BY-SA 4.0. The website's software licence is separate; no website code was imported. Modifications by name-za, 9 November 2026: converted word rows into JSONL, preserved grammar and original notes, trimmed surrounding whitespace and NFC-normalised candidate spellings while retaining originals, selected complete English-equivalent matches, and added provisional naming tags and review status. No translations were generated. Database examples and user-account tables were not imported. ## Expansion data On 10 November 2026, the same source snapshots were selected against an expanded editorial English concept list. The derived `data/imports/sa-depth-review.json` and `sa-depth-provenance.json`, and their copies in the live MySQL dictionary, are also **CC BY-SA 4.0**, with the source-specific credits above. Modifications: selected complete source gloss/equivalent matches and added provisional industry/style/concept tags. Original spellings, meanings, alternative senses, grammar, source notes and source/history URLs remain in provenance. No translations or missing grammar were generated. Preserve attribution, licence links and these modification notices when redistributing or adapting the dictionary data. The entire derived `data/imports/sa-review-expansion.json` and `sa-review-expansion-provenance.json` files are supplied under CC BY-SA 4.0, with the appropriate source-specific credits above and per-entry URLs. Preserve these notices and share adaptations under that licence. This is a statement about dictionary data, not a licence assignment to unrelated project code or original branding assets. Mafoko/NOODL and noncommercial terminology data has not been mixed into these files. ## Local naming selections The derived `data/quality/local-naming-words.json` selection (10 November 2026) retains the dictionary sources and recorded meanings for 400 whole words, with 40 selections in each of ten local languages. This selection and its editorial naming annotations are supplied under **CC BY-SA 4.0**. Modifications: selected individual existing entries, added business associations and AI assessment provenance, and preserved original spelling, meaning and source attribution. This does not certify fluent-speaker approval or infer grammatical combinations. Preserve the per-word credits and this notice when redistributing the selection. Its licence does not cover unrelated application code or branding assets. ## South African slang and regional senses Wiktionary contributors, via Kaikki.org / Wiktextract (Tatu Ylonen): [South-Africa tagged English senses](https://kaikki.org/dictionary/English/tags/4Y/South-Africa/index.html) and [South-African tagged English senses](https://kaikki.org/dictionary/English/tags/A7/South-African/index.html). Snapshot acquired 9 November 2026; the index identifies the 1 October 2026 English Wiktionary dump and 30 October extraction. See the original page and contributor-history links in every selected sense and the Wiktionary copyright policy saved with the first Kaikki acquisition. The derived `prepared/sa-slang/` data, `data/imports/sa-slang-review.json`, `sa-slang-provenance.json`, and those entries in the live dictionary are **CC BY-SA 4.0**: . Preserve attribution and share adaptations of this dictionary data under that licence. Modifications by name-za, 9 November 2026: selected regional senses, lowercased workshop spellings while retaining source case, added provisional naming tags and review notes, and stored alternative regional senses. No meanings or translations were generated. Quotations, examples, media and etymologies are excluded from derived data. Original downloaded snapshots remain separate; the text licence is not an assertion about third-party quotations or media in those originals. This collection contains no Urban Dictionary definitions.