រៀនបាលី — Pali for Khmer speakers

A Duolingo-style web app for learning Pali through Khmer, from the first letter to reading the Tipitaka, the commentaries and the Abhidhamma unaided.

Pali has no script of its own. In Cambodia it is written in Khmer script, and that is what a student actually meets in a ព្រះត្រៃបិដក — so every word here is shown in Khmer script and in IAST romanisation, and the learner chooses which.

Why it is built this way

It starts with what you already chant. Unit 3 — before any grammar — is នមោ តស្ស, សរណគមន៍ and សីល ៥. A student who already knows the sound of these gets their first experience of understanding Pali within about fifteen minutes. That is what brings someone back on day two.

Nothing is hand-authored twice. A word is stored once as a citation form; the morphology engine derives all sixteen noun forms and thirty verb forms from it. So the practice engine has an unlimited supply of drills, the grammar tables can never disagree with the exercises built from them, and adding one dictionary entry adds it to every drill type it can support.

Repetition is scheduled, not guessed. Every memorable thing — a letter, a word, a case ending, a line of a sutta — is one card in a spaced-repetition scheduler. Answer speed becomes the grade, so a hesitant correct answer comes back sooner than a confident one. As the review backlog grows, the intake of new cards automatically shrinks; drowning in reviews three weeks in is the usual reason people abandon an SRS app.

Participles come early. Absolutives and past participles (-tvā, -ta) arrive at unit 10, well before they would in a traditional syllabus, because they are what sutta prose is actually made of. Long sentences in the Dīgha Nikāya are chains of absolutives ending in one main verb; once you can see that, the Canon opens.

What is in it

Units / lessons	20 units, 81 lessons, five levels (មូលដ្ឋាន → កម្រិតខ្ពស់)
Dictionary	280 entries with Khmer and English glosses
Grammar	34 points in Khmer, using traditional terms (បឋមាវិភត្តិ, បុព្វកិរិយា…)
Reading	10 texts, glossed word by word with a full grammatical parse
Exercises	1,110 generated across the course, plus unlimited drill generation
Paradigms	15 noun classes, 5 tenses × 6 persons, participles, 8 pronoun tables

The reading corpus runs from នមោ តស្ស through Dhammapada 1 and 183, the Maṅgala and Mettā Suttas, and the first noble truth from the Dhammacakkappavattana Sutta, to a commentarial definition and the opening verse of the Abhidhammatthasaṅgaha.

Exercise types

Recognition and production multiple choice · Khmer-script ↔ romanisation · free typing (diacritics optional, Khmer script accepted) · listening · recitation aloud · sentence building from word tiles · cloze · grammatical parsing ("which case is ភិក្ខូនំ?") · generated declension and conjugation drills · metre drills (គរុ / លហុ).

All four skills are covered: reading through the glossed corpus, writing through free typing and the composition unit, listening through audio drills, and speaking through recitation.

Personalisation

• A placement check so an experienced reader is not made to grind the alphabet.
• Daily goal, new-cards-per-day, and script preference, all adjustable.
• Weak-area practice built from the tags you actually get wrong.
• Wrong answers return to the end of the session, so a lesson cannot be finished while something in it is still unlearned.

Running it

npm install
npm run dev        # development server
npm run build      # typecheck + production build to dist/
npm run preview    # serve the built site
npm test           # unit tests for the script, morphology and SRS engines

Deploying

The build is a static site, so it can be hosted anywhere.

Vercel (configured, and the simplest path): import the repository at vercel.com/new and accept the defaults — vercel.json already sets the framework, the build and install commands, the output directory, immutable caching for hashed assets, and the SPA fallback. Nothing needs to be configured in the dashboard, and there are no environment variables or secrets to set. Or from a terminal:

npm i -g vercel
vercel          # preview deployment
vercel --prod   # production

The app is entirely client-side, so no serverless functions or database are involved. One header worth knowing about: Permissions-Policy allows the microphone for this origin only, which the recitation drills need.

GitHub Pages is also supported, via the manually triggered Deploy to GitHub Pages workflow. It builds with BASE_PATH set to the repository name, since project pages are served from a subpath. Routing is hash-based precisely so that neither host needs rewrite rules.

Privacy

Everything is local-first. There is no account and no server: progress lives in the browser's storage, and can be exported to JSON and imported on another device. That keeps the app free to host, private by construction, and usable offline — which matters in a temple with poor connectivity.

A note on the audio and the microphone

No browser ships a Pali voice, and no engine recognises Pali. Listening exercises borrow the nearest available Indic voice, which renders Pali syllables far better than an English one but is still an approximation. Recitation drills will run speech recognition in a nearby locale where the browser offers it, but treat the transcript as a hint only — the grade is the learner's own judgement, and the app says so plainly.

Take your teacher's recitation as authoritative, not the synthesiser.

Licence and sources

The Pali texts are from the Pali Canon and are in the public domain. The Khmer translations, glosses and grammatical notes were written for this app and follow standard Khmer temple grammar terminology; corrections from teachers are very welcome.