BUSINESS · LANGUAGE TECH

Ijaw Language Curator

A collaborative platform where AI drafts Ijaw dictionary entries and the community corrects them into a verified, dialect-tagged dataset for language learning.

↓ SCROLL

The problem

Ijaw is spoken across the Niger Delta, but it has almost no structured digital record. Its dialects vary widely and existing dictionaries are thin, so learners and language tools have little reliable material to draw from. General AI models will happily produce Ijaw word meanings, but they are frequently wrong, and there is no trusted source to check them against. The result is a language that is hard to learn online and effectively invisible to modern software.

The solution

Ijaw Language Curator turns dataset building into a community effort. Gemini drafts candidate entries with word, meaning, pronunciation, and dialect, then members review each one and submit corrections that others vote to approve or reject. Every verified entry feeds a clean, dialect-tagged dataset for language learning. Points, streaks, achievements, and a leaderboard keep contributors coming back, while an admin queue holds quality high. Members can record their own pronunciations, hear text-to-speech playback, and type Ijaw characters with a built-in virtual keyboard.

What it delivers

AI-seeded entries

Gemini generates candidate words and full sentences with meaning, pronunciation, and dialect, so the dictionary starts from a draft instead of a blank page. Difficulty levels and exclusion lists keep batches fresh.

Community correction and voting

Members submit corrections to any entry and vote to agree or reject, so meanings converge on what speakers actually use. Entries move from pending to verified or flagged as the community weighs in.

Gamified contribution

Points, daily streaks, achievements, and a leaderboard reward steady participation, with challenges tied to contributions, streaks, and friends to sustain momentum.

Voice and pronunciation

Contributors record their own audio for words and play back text-to-speech pronunciations, capturing how each dialect actually sounds rather than only how it is spelled.

Dialect-aware, moderated dataset

Every word and correction is tagged by dialect and moderated through an admin panel, producing a high-quality dataset ready to power language learning and downstream tools.

Built with

React 19TypeScriptGemini APIFirebaseFirestoreViteTailwind CSSFramer MotionThree.jsExpress
2026Year
SoloRole
Gemini + FirebaseStack
Community-curatedModel