How it works

From a typed word to a globe path — what Lexicon Earth fetches, what it invents, and how to read the result.

1. Search

You enter a word in any script Wiktionary supports (for example English, Spanish, Arabic, Chinese, Hindi, or Russian). The app looks up the matching page — trying common capitalizations for Latin scripts, Unicode normalization, and OpenSearch / fulltext when needed — then reads the language sections on that page. When your query’s script suggests a language (Arabic script → Arabic/Persian, Cyrillic → Russian, and similar), that section is preferred.

If several etymologies or language entries exist, you can choose which sense to load in the panel. Optional multi-source mode can also query Wikidata and Kaikki.

2. Parse a lineage

Etymology templates in Wiktionary (inheritance, borrowing, derivation, and related forms) are parsed into a chain of word-forms. Each form keeps its language label, gloss when available, and epistemic status (attested, reconstructed, disputed).

Extracted prose summaries are labeled as extracted text — not quotations of scholarly consensus.

3. Place forms on the map

Languages are mapped to approximate heartlands (a point for the language’s historical or modern center). That is an association for visualization, not evidence that a specific inscription or speaker lived at that coordinate.

When a language code is missing from our table, that form may appear in the panel without a map marker. The globe bridges mapped neighbors so the journey still reads as a path.

4. Read the journey

Example: Searching orange may show a chain through India, Persia, Arabia, and Western Europe. The arcs illustrate sequence in the extracted etymology — not a documented travel route.

5. Cache & offline

Successful lookups are stored locally. A small starter set ships with the app. After the first visit, a service worker can keep the shell, map data, and libraries available offline. New words still need the network.

Limits worth knowing

Last updated: 13 October 2026