Liturgical Armenian Transliteration

Convert Classical Armenian (Grabar) Unicode text to English transliteration using Western Armenian liturgical pronunciation conventions.

Available as an npm package — works in Node.js (CJS & ESM) and the browser.

Installation

npm install liturgical-armenian-translit

Usage

Node.js (CommonJS)

const { transliterate, transliterateWord, loadDictionary } = require('liturgical-armenian-translit');

transliterateWord('Աստուած');
// => "Asdvadz"

transliterate('Տէր Աստուած մեր, կեցո՛ զմեզ եւ ողորմեա՛։');
// => "Der Asdvadz mer, getzo uzmez yev voghormia:"

Node.js (ESM)

import { transliterate, transliterateWord } from 'liturgical-armenian-translit';

transliterateWord('սուրբ');
// => "soorp"

Browser (CDN via importmap or script tag)

<script type="module">
import { transliterate, transliterateWord } from 'https://cdn.jsdelivr.net/npm/liturgical-armenian-translit@1/dist/translit.browser.mjs';

console.log(transliterate('Տէր Յիսուս Քրիստոս'));
// => "Der Hisoos Krisdos"
</script>

Custom dictionary entries at runtime

loadDictionary({ Մահր: 'Mahr', գետ: 'get' });
transliterateWord('Մահր');
// => "Mahr"

Bundled dictionary (JSON)

const dictionary = require('liturgical-armenian-translit/src/dictionary.json');
console.log(dictionary['սուրբ']);
// => "soorp"

API

transliterate(text)

Transliterate full Armenian text while leaving non-Armenian content unchanged.

transliterateWord(word)

Transliterate a single word using exact lookup, case-insensitive lookup, stem lookup, then rule-based fallback.

transliterateByRules(word)

Apply only the fallback phonology rules.

loadDictionary(dictData)

Merge custom Armenian-to-Latin mappings into the in-memory dictionary at runtime.

Accuracy

The repository includes a normalized Divine Liturgy corpus at corpus/divine-liturgy.json. Running npm run test:corpus currently measures:

| Metric | Value | | --- | --- | | Corpus segments | 220 | | Paragraph exact match | 100.0% | | Word exact accuracy | 99.8% | | Word mismatches | 7 out of 3,430 |

A baseline snapshot is tracked at test/fixtures/corpus-metrics-baseline.json; npm test fails if exact-match counts regress.

Development

• npm test runs API and regression tests with Node's built-in test runner.
• npm run test:corpus prints corpus-level evaluation metrics using sequence-aligned word comparison.
• npm run metrics prints the checked-in corpus metric snapshot in JSON form.
• npm run build:browser generates dist/translit.browser.mjs — a self-contained ESM build for the browser.
• npm run build:dictionary generates derived dictionary artifacts under corpus/derived/ from the checked-in corpus.
• npm run analyze prints character frequencies, first-letter mappings, and mismatch buckets.
• npm run audit:dictionary writes a dictionary quality report to corpus/analysis/dictionary-audit.json.

The browser build is regenerated automatically via prepublishOnly before each npm publish.

Limitations

• Optimized for liturgical Classical Armenian, not modern spoken Eastern or Western Armenian.
• Morphology handling is heuristic and intentionally lightweight.
• Proper nouns and rare forms may still need custom dictionary entries.

License

MIT