Mojido (文字道)

The Way of Characters - A Japanese reading practice app that teaches you to read Japanese by typing romaji from memory.

How It Works

1. See Japanese - A sentence appears with the current token highlighted
2. Recall the reading - You must remember the romaji (no hints shown by default)
3. Type it - Your typed romaji appears below with instant feedback
4. Learn from mistakes - Errors block progress until you backspace and correct them

This is active recall practice - you're training your brain to recognize and read Japanese characters, not just copy what you see.

Features

• Real sentences from Tatoeba - learn with meaningful context, never random characters
• Token-by-token validation with instant green/red feedback
• Kanji hints - click any kanji (dotted underline) to reveal its reading
• Smart romaji handling:
  • Multiple input styles accepted (し = "shi" or "si", つ = "tsu" or "tu")
  • Proper ん handling (requires "nn" before vowels/y-sounds)
  • っ (small tsu) consonant doubling
• Day streak tracking with celebration on first sentence each day
• Progress tracking - per-character accuracy stored locally
• Dark mode support

Screenshots

Coming soon

Installation

Prerequisites

• Bun (package manager)
• Rust (for Tauri)

Development

# Install dependencies
bun install

# Run development server
bun run tauri:dev

# Or use axogen
bunx axogen run dev

Build

# Build for production
bun run tauri:build

# Or use axogen
bunx axogen run build

Tech Stack

• Frontend: Svelte 5, SvelteKit, TailwindCSS v4
• UI Components: shadcn-svelte
• Desktop: Tauri v2
• Database: SQLite (via tauri-plugin-sql)
• i18n: Paraglide

Data Sources

• Sentences: Tatoeba (CC BY 2.0 FR)
• Tokenization: Kuroshiro + kuromoji

Generating Sentences

The app comes with pre-generated sentences. To regenerate:

bunx axogen run generate

This downloads Japanese sentences from Tatoeba, tokenizes them with readings, and outputs to src/lib/data/sentences-generated.ts.

License

MIT License - see LICENSE for details.

Acknowledgments

• Tatoeba contributors for the sentence corpus
• The Kuroshiro project for Japanese text processing

---

Built with ❤️ by Oliver Seifert