whitakers-words

A modern TypeScript port of Whitaker's Words, a comprehensive Latin dictionary and morphological analyzer originally written in Ada by William Whitaker.

Try the online demo

Given any Latin word form, the engine strips inflectional endings, matches stems against a 39,000-entry dictionary, and returns every possible parse: part of speech, declension/conjugation, case, number, gender, tense, voice, mood, and English meaning. It also supports English-to-Latin reverse lookup, enclitic/prefix/suffix stripping, spelling tricks for medieval and variant orthography, and unique irregular forms.

Installation

npm install whitakers-words

Quick Start

Node / Bun / Deno

The whitakers-words/node entry point loads the bundled data files from disk automatically:

import { createEngine, formatWordAnalysis } from "whitakers-words/node";

const engine = createEngine();

// Parse a Latin word
const analysis = engine.parseWord("amare");
console.log(formatWordAnalysis(analysis));
// am.are               V      1 1 PRES ACTIVE  INF 0 X
// amo, amare, amavi, amatus  V (1st)   [XXXAO]
// love, like; fall in love with; be fond of; have a tendency to;
// ...

// English-to-Latin reverse lookup
const results = engine.searchEnglish("water");
// => [{ de: { stems: ["aqu", ...], mean: "water; sea, lake; ..." }, rank: 6 }, ...]

Browser

The main whitakers-words entry point is platform-agnostic (no fs or path imports). You provide the data file contents yourself: typically via fetch:

import { WordsEngine, formatWordAnalysis } from "whitakers-words";

const load = (url: string) => fetch(url).then((r) => r.text());

const [dictGen, dictSup, inflects, addons, uniques] = await Promise.all([
  load("/data/DICTLINE.GEN"),
  load("/data/DICTLINE.SUP"),
  load("/data/INFLECTS.LAT"),
  load("/data/ADDONS.LAT"),
  load("/data/UNIQUES.LAT"),
]);
const dictline = `${dictGen}\n${dictSup}`;

const engine = WordsEngine.create({ dictline, inflects, addons, uniques });
console.log(engine.formatWord("aquam"));

The five data files are included in the npm package under data/. Concatenate DICTLINE.GEN and DICTLINE.SUP into a single dictline string (the Node entry point does this automatically). How you serve them is up to your setup (static assets, CDN, bundler plugin, etc.).

CLI

The package includes a whitakers command:

# Parse Latin words directly
npx whitakers "rem acu tetigisti"

# Interactive mode
npx whitakers
=> aquam
aqu.am               N      1 1 ACC S C
aqua, aquae  N (1st) F   [XXXAO]
water; sea, lake; river, stream; rain, rainfall (pl.), rainwater; spa; urine;

# English-to-Latin mode
=> ~E
=> water
aqua, aquae  N (1st) F   [XXXAO]
water; sea, lake; river, stream; ...

# Switch back to Latin mode
=> ~L

Pipe-friendly: reads from stdin when not a TTY:

echo "saucia cura" | npx whitakers

API Reference

WordsEngine

The main class. Created via WordsEngine.create(data) or the convenience createEngine() from whitakers-words/node.

Method / Property	Description
static create(data: WordsEngineData)	Create an engine from raw data file contents
parseWord(word: string, nextWord?: string): WordAnalysis	Analyze a Latin word with optional lookahead for compound verbs
searchEnglish(word: string, maxResults?: number): EnglishSearchResult[]	English-to-Latin reverse lookup
parseLine(line: string): WordAnalysis[]	Parse a full line of Latin, detecting compound verbs across words
formatLine(line: string): string	Parse and format a full line as human-readable text
formatWord(word: string): string	Parse and format a single word as human-readable text
dictionarySize: number	Number of dictionary entries (expect ~39,000)
inflectionCount: number	Number of inflection rules (expect ~1,800)
uniqueCount: number	Number of unique/irregular entries (expect ~76)
addons: AddonsData	The parsed addons (prefixes, suffixes, tackons)
trimOutput: boolean	Ada's Trim_Output (default true); false keeps archaic and uncommon inflections

WordAnalysis

Returned by parseWord(). Contains all possible interpretations of a word:

以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗