A growing collection with visible sources, uneven coverage, and room for knowledgeable speakers to help it improve.
The main entries are adapted from community-created Wiktionary dictionaries using structured extracts from Kaikki / Wiktextract. Headwords, definitions, grammatical labels, pronunciation notation, forms and selected examples are retained when present. Entries have not been independently reviewed by Jeliya language editors.
English, French, Spanish and Portuguese source editions have different coverage. In addition, a small basic vocabulary supplies draft French, Spanish and Portuguese translations of explicit English senses. Those are labelled on individual entries and have not received native-speaker or professional editorial review. They are not represented as translations supplied by the original source.
Some Manding records are Jeliya editorial adaptations used by permission. Their source URLs remain available for verification; these records do not automatically inherit the Wiktionary licence.
This is an initial collection, not a database of every African language or an exhaustive dictionary. Each language has a multi-year study roadmap; vocabulary exercises and verified resources vary in depth. Broad labels such as Fula, Akan and Arabic need a regional or standard variety to become a focused course. Haitian Creole is included separately for the diaspora. Countries shown in the directory are illustrative, not exhaustive.
Wiktionary contributors are credited through each entry’s original page and contributor history. The adapted data is shared under Creative Commons Attribution-ShareAlike 4.0; consult Wiktionary’s copyright terms for source licensing. Changes include selection, format conversion, selected example retention, search indexes, and the explicitly labelled draft gloss translations.
Extraction reference: Tatu Ylonen, “Wiktextract: Wiktionary as Machine-Readable Structured Data,” LREC 2022. The coverage manifest records original source pages, available download links, retrieval dates and entry counts. Full entries are accessible in the dictionary; larger editions load detail files on demand.
The 48-module roadmaps combine language-specific structural maps, original study tasks and linked resources. They are CEFR-informed plans, not accredited courses or guaranteed C2 outcomes. Some maps are necessarily provisional at the variety level; learners need local speaker feedback. Teacher profiles describe what teachers report about themselves; the platform does not label these credentials as verified.
Jeliya Accounts keep the name and email supplied at registration together with saved words, course progress, SRS history, XP, streaks, badges, teacher-profile information, lesson requests, and subscription status where those features are used. Pronunciation contributions are stored separately and remain private until approved. Recovery codes are shown once; password-reset email and email ownership verification are not enabled in this release.
Learning plans are educational guidance, not accredited qualifications or guarantees of proficiency. Dictionary sources retain their stated licences. Community submissions may remain pending or be rejected during review. Teacher profiles are self-reported, and learners and teachers arrange lesson format, availability, fees, and payment independently unless a future checkout explicitly says otherwise.
Entry counts include distinct headword / word-type records and draft gloss translations; they are not counts of unique concepts. Zero means that edition is unavailable.
Help grow this collection through teaching, language expertise, and careful source-based correction. A living language deserves more than a single definition.