Il-Miġma’

Il-Miġma’ is an AI-assisted Maltese lexical platform that combines a dictionary, morphology engine, root explorer, and learning workspace into one research-grade interface.

Il-Miġma’ ("the collection") is built for people who need more than basic definitions: linguists, language learners, educators, lexicographers, and developers building Maltese language tools.

Who it is for

• Learners who want dictionary entries with morphology, roots, IPA, and usage context.
• Researchers and linguists who need searchable Semitic root structure and derivational patterns.
• Teachers and course builders who need structured lexical data for drills and learning content.
• Contributors and editors who need an admin workflow for curating entries, roots, and suggestions.
• Developers who want to reuse lexical search and structured morphology through API endpoints.

Core features

• Dictionary search with exact, full-text, gloss, and word-form dimensions.
• Root explorer for Semitic consonantal roots, patterns (wiżen), and derived forms.
• Conjugation engine that uses root class + vowel sets for generated verb paradigms.
• Browse mode for discovery-based exploration beyond direct search.
• Learning surfaces such as course/blog pages and scaffolded lexical navigation.
• AI chatbot assistant for guided lexical lookup and user support.
• Admin workspace for content curation, relationships, and database tooling.
• Suggestion pipeline for community contribution and moderation.

Product surfaces at a glance

• /search — dictionary lookup
• /root-search and /root/:id — root and pattern exploration
• /conjugator — morphology/conjugation exploration
• /chatbot — AI assistant surface
• /browse — exploratory lexical browsing
• /admin — editorial + data maintenance workspace
• /suggest — suggestion intake

Stack

• Frontend: React 19 + TypeScript + Vite + React Router + Tailwind CSS.
• Auth: Clerk.
• Backend: Cloudflare Pages Functions (functions/api/*).
• Database: Turso (libSQL / SQLite-compatible schema).
• AI integration: Gemini via @google/generative-ai.
• Storage: Cloudflare R2 integration support for audio and media workflows.

Data model highlights

The linguistic model separates abstract morphology from surface forms:

• roots store consonantal skeletons (e.g. k-t-b) and class metadata.
• patterns store CV/wiżen templates.
• root_pattern_forms join roots + patterns into realizations.
• entries store the actual dictionary headwords and lexical attributes.
• definitions, phonetics, etymologies, dialect_variants layer meaning and evidence.

This makes Il-Miġma’ both searchable for end users and reusable as structured lexical infrastructure.

For contributor-level docs, see:

• docs/data-model.md
• docs/import-pipeline.md
• docs/search-api.md

Screenshots / demo tour

Recommended: replace these with GIFs as workflows stabilize.

Surface	Preview
Home	public/screenshots/home.png
Search	public/screenshots/search.png
Root Search	public/screenshots/root-search.png
Conjugator	public/screenshots/conjugator.png
Chatbot	public/screenshots/chatbot.png
Browse	public/screenshots/browse.png
Admin	public/screenshots/admin.png
Suggest	public/screenshots/suggest.png

Local setup

1) Install dependencies

npm install

2) Configure environment

Create .env in the project root:

VITE_TURSO_URL=libsql://<your-db>.turso.io
VITE_TURSO_AUTH_TOKEN=<token>
VITE_CLERK_PUBLISHABLE_KEY=<clerk_publishable_key>
GEMINI_API_KEY=<gemini_key_if_using_chat>

For wrangler pages dev (npm run dev:api), copy .dev.vars.example to .dev.vars and set Worker runtime vars:

TURSO_URL=libsql://<your-db>.turso.io
TURSO_AUTH_TOKEN=<token>
CLERK_SECRET_KEY=dummy

file:// database URLs are not supported by the Worker runtime. Use remote libsql:// or https:// when running the API locally with Wrangler.

3) Run the frontend

npm run dev

4) Run Cloudflare Pages Functions locally

npm run dev:api

5) Optional: import lexical data

npm run import:dry
npm run import:all

Docs index

• docs/README.md — full docs index
• docs/data-model.md — contributor-oriented schema and relationships
• docs/import-pipeline.md — spreadsheet → DB pipeline
• docs/search-api.md — /api/search parameters and examples