LangDex

Language Learning Apps

Every language app ships the same JMDict/CC-CEDICT data because there's no alternative. LangDex provides the graph layer that turns isolated dictionary entries into traversable knowledge structures—with proficiency-aware filtering built in.

The Challenge

The Dictionary Wrapper Problem

  • 1Your app is architecturally identical to competitors: flat JSON files from JMDict, Kaikki, or CC-CEDICT with a UI layer
  • 2No semantic links between homophones, synonyms, or kanji—users can't explore why 上げる and 挙げる differ
  • 3Generating intelligent quiz distractors requires manual curation; the data doesn't encode semantic similarity
  • 4Proficiency filtering (JLPT N5 vs N1) requires maintaining separate word lists instead of querying a tagged graph
  • 5Shipping 500MB+ SQLite databases for offline mode bloats app size and slows cold starts
The Solution

Graph-Native Learning Infrastructure

LangDex's Learn lens projects the linguistic graph into learner-optimized response shapes. Each lexeme links to PanLex concepts (enabling instant cross-lingual lookup), Tatoeba sentence IDs (10M+ examples), KanjiDic2 character data, and proficiency tags (JLPT N5-N1, HSK 1-6). Query once, get the full learning context—no joins, no secondary lookups.

Features

Key Capabilities

Proficiency-Tagged Lexemes

Every Japanese lexeme tagged with JLPT level (N5-N1), Chinese with HSK (1-6). Filter API responses by learner stage without maintaining separate word lists.

Concept-Based Traversal

Click a word, traverse to its PanLex concept, retrieve translations in 5,700+ languages. Click a kanji, get all KanjiDic2 readings and compounds.

Semantic Distractor Generation

Generate quiz options from the graph: words sharing radicals (視/見/観), phonetic similarity, or semantic proximity. No manual curation required.

Pitch Accent & IPA

Japanese pitch accent patterns from accent dictionaries. 4.7M IPA transcriptions from IPA-Dict across languages.

Compressed Offline Packs

Download language-specific graph slices as compressed binaries. Delta sync keeps local data fresh without re-downloading.

Tatoeba Sentence Links

Each sense links to example sentences from Tatoeba's 10M+ corpus. Parallel translations available for 400+ language pairs.

Who Is This For

Perfect For

Japanese dictionary apps (Takoboto, Shirabe Jisho competitors)Chinese learning platforms (Pleco alternatives)SRS/flashcard tools (Anki add-on developers)Immersion readers (Yomitan, Language Reactor)Browser extensionsLanguage game developersEdTech startupsIndie developers tired of parsing raw JMDict XML

Ready to Get Started?

Join our waitlist to get early access to LangDex's API and start building.