About

About Kaidoku 解読

解読 (kaidoku) means “deciphering”. Kaidoku is an open-source companion for reading Japanese at JLPT N5–N3: it breaks down the grammar in any text, and teaches the kanji and vocabulary you need to read it.

Three modules

Grammar Pattern Breakdown: A rule-based analyzer, not an AI model: 341 hand-written grammar rules run over the output of the Kuromoji morphological analyzer, so the same text always gets the same result. It runs on the server and needs an internet connection.
Kanji Learning: 640 kanji in study sets of ten, most common first, with study cards, reading quizzes, meaning checks and spaced review. Works offline once you have opened it.
Vocabulary Learning: 2,963 words in study sets of about fifteen per level, with the same cards, quizzes and review. Works offline once you have opened it.

Learning data and its licence

Contains data from OpenJLPT (CC BY-SA 4.0), which uses JMdict and KANJIDIC2 (EDRDG), Jonathan Waller's JLPT lists, and Tatoeba.

The kanji and vocabulary data is a derived work of OpenJLPT, pinned to revision 0d1d3410bec90bd4098a7c72de820543cb4f707c, and is shared under CC BY-SA 4.0. Read the licence text and the upstream notice.
Readings and meanings are exactly as given by these sources. They have not been independently edited by a lexicographer.
Example sentences are not shown, and meanings are self-checked rather than multiple choice, because both are still waiting for an independent bilingual review.
The JLPT does not publish official vocabulary or kanji lists. Level assignments are community estimates based on Jonathan Waller's lists and are not an official guarantee. Grammar levels follow commonly used N5–N3 study lists for the same reason.

Your progress stays with you

There are no accounts, analytics, cookies or advertising.
Quiz answers and review schedules are stored only in this browser (IndexedDB). They never leave your device unless you download a backup yourself.
A missed item comes back after ten minutes; a correct one after a day, then at growing intervals of up to a year. Each reading or meaning track counts as learned after two correct answers in a row, including practice.
Text you paste into the grammar analyzer is sent to the server only to be analysed. It is not stored.

Source code and software

Kaidoku's source code is open source under the MIT licence: github.com/froxily/kaidoku.
It is built with Spring Boot (Apache 2.0) and Kuromoji with the IPADIC dictionary (Apache 2.0; the dictionary carries its own permissive licence). The full list is in the repository's THIRD_PARTY_NOTICES file.