Kanji Quiz
Practice kanji as active recall instead of a thin flash-card page: type meanings, produce readings, reverse English prompts and cycle real misses through local review.
Meaning recall is typed, not guessed
The main Meaning 10 mode shows one kanji and asks you to type an accepted English meaning. That matters because recognizing the right answer among four buttons is easier than producing an answer from memory. If the first response is wrong, the game records the miss and gives one clean retry before revealing the source-backed meanings. The round score is based on first-attempt accuracy, so recovery is useful for learning without pretending a second-try answer was equivalent to immediate recall.
Every kanji in the runtime bank comes from a frozen source snapshot rather than an improvised hand-written list. The current bank contains 2,211 unique characters grouped into N5, N4, N3, N2 and N1 study bands. Those level labels come from community data, not an official JLPT specification. You can select one band for focused practice or use the All option when you want a broad search/reference view.
Reading practice accepts kana or common romaji
Reading 10 starts from the kanji itself and asks for any listed on-reading or kun-reading. Answers can be typed in Japanese kana or common romaji. Source punctuation such as a dot used to show okurigana boundaries is normalized for answer checking, so the exercise tests the reading rather than whether you reproduced dictionary formatting.
Kanji can have several legitimate readings, and the game does not collapse them to one artificial “correct” sound. If a listed on or kun reading is present in the frozen dataset, it is accepted. The result feedback shows the English meaning plus compact on/kun context so a correct response still teaches something beyond a green checkmark.
Reverse recall avoids ambiguous English prompts
Reverse 10 begins with an English meaning and asks you to choose the matching kanji. A naive reverse quiz can be unfair because many kanji share broad English glosses such as “person,” “use,” “go,” or “old.” This engine therefore builds its reverse pool only from primary meanings that are unique within the selected level. If a prompt would have more than one defensible answer in that pool, it is not used as an ordinary reverse question.
This is also why the site does not create hundreds of separate pages for individual levels, scores or settings. N5 through N1 are real filters inside one engine. Search-facing pages are reserved for genuinely different user intent, while configuration stays in the game where it belongs.
Weak Kanji and visible review intervals
A kanji you actually miss enters Weak Kanji immediately. Correct first attempts progress through visible local review stages of 1, 3, 7, 14 and 30 days. The review schedule is intentionally transparent: it is a browser-local interval system, not a claim that the site has calculated your memory strength, learning probability or official proficiency.
Due + Weak Review combines scheduled items and actual misses. Three correct Review attempts retire an item from the weak queue. Your practice history, due dates, Sprint bests and Daily streak stay on this browser. No login, backend profile or paid learning API is required.
Searchable Kanji Bank with source context
Kanji Bank is the reference side of the same product. Search by character, English meaning, on-reading or kun-reading. Cards show the selected study level, stroke count, source frequency when available, grade when available, readings, meanings and your own local practice history. Browsing the bank does not increment “seen,” alter due dates or create a fake mastery record.
The data provenance is explicit. The frozen release uses the OpenJLPT derivative dataset, whose level assignments are attributed upstream to Jonathan Waller community lists and whose dictionary details are derived from KANJIDIC2 from the Electronic Dictionary Research and Development Group. The derivative dataset is distributed under CC BY-SA 4.0, and the Project archive retains the five source CSV files, their hashes and attribution information.
JLPT labels are study groupings, not certification
The Japanese-Language Proficiency Test does not publish a definitive post-2010 vocabulary or kanji specification. That means it would be misleading to call any third-party N5–N1 list “the official JLPT kanji list.” This game does not make that claim. The level filter is useful for organizing practice because the underlying community assignments are widely used study approximations, but a quiz score here is not a JLPT score and does not certify mastery.
The practical learning path on this platform is Kana first, then useful Vocabulary, then Kanji. Each specialist keeps its own review and Daily history so a learner can work at the right layer without forcing language study into the six-game brain-puzzle Daily Score.