Guide · Anki card design for Japanese

Monolingual Anki kanji generation cards

The phrase packs two separate card-design decisions into one label, and most explanations only cover one of them. This is the version that pulls the two apart: the recall direction that makes a card a generation card, the language choice that makes it monolingual, and why stacking both is the strongest, most demanding kanji card you can build.

Matthew Diakonov, Written with AI

Published May 17, 20268 min read

Direct answer

A prompt with no English, and you produce the kanji.

A monolingual Anki kanji generation card is a card where the prompt side carries only Japanese, a Japanese-to-Japanese dictionary definition or an example sentence with the target word removed, and your job is to produce the kanji yourself rather than recognize it. Two design choices are stacked. Generation sets the direction: the answer is the kanji, the prompt is not. Monolingual sets the language: there is no English keyword anywhere on the card.

Each choice closes a different shortcut. Generation removes the option of passively recognizing a character you would never be able to summon on your own. Monolingual removes the option of pattern-matching an English gloss instead of actually understanding the Japanese. A card with both closed is the most honest test of kanji knowledge Anki can hold, and also the most expensive in review effort, which is why it is a tool you aim selectively, not a default for a whole deck.

The two axes, and why they get confused

Most guides treat "monolingual cards" and "production cards" as two separate topics in two separate articles. They are not separate. They are two independent axes, and any kanji card sits somewhere on both at once. The label in the heading above is just the name for one specific corner of the grid.

	Bilingual prompt (English present)	Monolingual prompt (Japanese only)
Recognition direction see the kanji, recall meaning	Kanji on the front, English keyword on the back. The easiest card and the right starting point. You can pass it by matching a familiar shape to a familiar word.	Kanji on the front, J-J definition on the back. Still recognition, but reading the definition forces real comprehension instead of an English label.
Generation direction produce the kanji from a prompt	English keyword on the front, kanji on the back. Production, but the prompt is an English word, so you risk learning a translation pair rather than the Japanese word itself.	J-J definition or sentence on the front, kanji on the back. Both shortcuts closed. This is the monolingual kanji generation card.

The grid also explains the order people are told to move through it. You start in the top-left corner because it is the only corner a beginner can actually use. As reading improves, the monolingual column opens up. As you start needing to write and speak rather than only read, the generation row opens up. The bottom-right corner is where those two paths meet, which is exactly why it gets framed as an endgame rather than a default.

The cheat each axis closes

The reason this card type is worth the extra effort is specific, and it is the part most explanations skip. Easy cards are not easy because the material is easy. They are easy because they leave a shortcut open, and the brain takes the shortcut every single time.

On a bilingual recognition card, the shortcut is shape matching. After a few reviews you are not recalling a meaning, you are recognizing a familiar silhouette and firing the English word attached to it. The card keeps passing and you feel like you know the kanji, right up until you see it in a sentence where the surrounding context is unfamiliar and the silhouette no longer triggers anything.

Switching to the generation direction closes that shortcut. There is no shape on the front to recognize, so passive familiarity earns you nothing. You either reconstruct the characters or you fail. Switching to a monolingual prompt closes a second, quieter shortcut: matching the first few words of an English gloss. A J-J definition gives your brain nothing in its native language to latch onto, so it has to process the Japanese to get anywhere.

This is not a Japanese-only idea. Any spaced-repetition system that lets you re-see the identical prompt is quietly training you to pattern-match the prompt rather than know the answer. Studyly, a coursework study tool unrelated to Japanese, was built around the same observation and states it plainly in its own product notes: "Auto-rephrasing means I can't lazy-pattern-match the first three words." Different subject, identical failure mode. A monolingual generation card is the manual, Japanese-specific way to shut that door.

Building one that pins exactly one answer

The single most common way a monolingual generation card fails is a prompt so loose that several kanji words would satisfy it. When that happens you grade yourself on a vague feeling, the card stops testing anything, and the effort is wasted. The build steps below all serve one goal: make the prompt point at one and only one word.

Choose the source for the prompt side

Pull a definition from a monolingual (Japanese-to-Japanese) dictionary, or take a real example sentence and delete the target word. A definition tests the concept; a sentence tests the word in context. Sentences pin a single answer more reliably, so prefer them when a definition feels broad.

Decide how much help the prompt gives

Including the reading on the front constrains the answer to one word and makes the card a pure writing test. Leaving the reading off makes it harder, because now you must produce both the reading and the characters. Pick one deliberately per card rather than drifting between them.

Swap the template fields

A mined card arrives as recognition by default. To make it a generation card, set the prompt side to the Japanese definition or sentence and the answer side to the kanji. In Anki this is a one-time note-type or card-template change, not per-card work.

Put furigana and audio on the answer side

A Japanese support add-on can attach furigana automatically. You need it on the answer so you can self-check the reading honestly. Audio on the answer side lets the card double as a listening anchor.

Grade strictly, and only ever as production

Pass only if you produced the exact kanji and reading the card wants. Approximate recall is a fail by design. If you keep passing on a vague sense of the word, the prompt is too loose: go back and add or tighten the example sentence.

A loose card and a tight card, side by side

Toggle between the two states below. Same target word, two prompts. One could be answered with half a dozen different kanji words; the other can only be answered one way.

Same target word, two prompt designs

Front: a one-line J-J definition along the lines of 'a feeling of wanting something you do not have.' Several distinct words fit that description. You look at it, think 'yes, something like that,' flip the card, see the kanji, and tell yourself you knew it. You did not. You recognized it after the fact. The card passed and taught nothing.

Multiple words satisfy the prompt
Self-grading drifts to a vague feeling
Recognition disguised as production
Passes easily, retains poorly

When the technique is actually worth it

Because each monolingual generation card costs two to three times the review effort of a plain recognition card, the question is never "are these good cards" but "which words deserve one." The checklist below is the filter worth applying before you convert anything.

Convert a word to a monolingual generation card when

You can already read intermediate Japanese well enough to parse a J-J definition without reaching for English.
You genuinely need to produce the word, in writing or speech, not only recognize it while reading.
The word lives in a near-synonym cluster where you keep confusing which kanji belongs to which meaning.
You have a real example sentence available, so the prompt can pin a single answer.
Skip conversion for words you only ever need to read, and for anything below your comfortable reading level.

Used this way, monolingual generation cards stay a small, sharp minority of a deck. The bulk of your reviews remain recognition cards, which carry coverage cheaply, and the generation cards do the narrow, expensive job of forcing real production on the words that warrant it.

Where Studyly fits, and where it does not

Studyly is not a Japanese kanji tool. It is honest about that.

If you arrived here from a study-tools thread, here is the straight answer: Studyly does not build J-J kanji decks, and you should not try to make it one. It turns lecture slides, PDFs, and textbooks into practice questions for memorization-heavy coursework, and its question-quality benchmark was measured on medical lecture documents, not Japanese material. For a kanji workflow, a pop-up dictionary feeding mined cards into Anki is the right path.

It earns a mention only because it was built around the same principle this page is about: a question you can pattern-match is a question that has stopped testing you. If your memorization load is professional coursework rather than a language, in medicine, dentistry, nursing, pharmacy, veterinary, or biology, that is the problem Studyly was made for.

Questions people ask about this card type

What is the difference between a recognition card and a generation card?

Recognition (also called recall-from-character) shows you the kanji or word and asks for its meaning or reading. Generation (also called production) does the reverse: the prompt is not the kanji, and you have to produce the kanji yourself, ideally by writing it or at least by picturing the exact characters. Recognition tests whether you can decode something you see. Generation tests whether you can summon it cold. They are different skills, and being good at one does not make you good at the other. Most learners are far stronger at recognition because that is what reading practice trains by default.

Do I need to be advanced before I use monolingual cards?

You need to be able to read intermediate Japanese comfortably, because a monolingual card replaces the English gloss with a Japanese-to-Japanese (J-J) definition or an example sentence. If you cannot parse the definition, the card teaches nothing. A common waypoint is roughly the upper-beginner to lower-intermediate boundary, often described as being able to read a definition written for children or early teens. Below that level, bilingual cards are not a crutch, they are the correct tool. Monolingual is an endgame technique, not a beginner one.

Should I make all of my kanji cards monolingual generation cards?

No. A monolingual generation card is roughly two to three times harder than a bilingual recognition card, and a deck made entirely of them burns review stamina fast. The usual approach is to keep the bulk of a deck as recognition cards and reserve generation cards for the specific words you genuinely need to produce: words you want in active writing or speech, or near-synonym clusters where you keep mixing up which kanji goes where. Convert selectively, not wholesale.

How do I grade a monolingual generation card fairly?

The pass bar for a generation card is stricter than for recognition. You pass only if you produced the exact kanji the card wants, including the correct characters and the correct reading. "I knew roughly which word it was" is a fail, because the entire point of the production direction is that approximate knowledge is not enough. Adding furigana on the answer side via a Japanese support add-on lets you self-check the reading. If you find yourself passing cards on a vague feeling, the prompt is too loose and needs an example sentence to pin it.

Can I auto-generate monolingual kanji cards from Japanese text?

Partly. Pop-up dictionary browser extensions and text-mining scripts can pull a word, its reading, and a J-J definition out of a passage and push them into Anki, which removes most of the typing. What no generator decides for you is the card direction. A mining tool drops a recognition card by default. Turning it into a generation card is a template choice you make once: swap the front and back fields so the prompt side carries the Japanese definition or sentence and the answer side carries the kanji. The automation handles the data; you handle the direction.

Is Studyly a tool for making Japanese kanji cards?

No, and it is worth being direct about that. Studyly converts lecture slides, PDFs, and textbooks into practice questions for memorization-heavy coursework: medicine, dentistry, nursing, pharmacy, veterinary, biology. Its question-quality benchmark was measured on medical lecture documents, not on Japanese-language material, so it is not the right tool for building a J-J kanji deck. It appears on this page only because it was built around the same idea this technique relies on, defeating pattern-matching, applied to a different subject. If you are studying Japanese, a pop-up dictionary plus a mining workflow into Anki is your path.

Anki itself does not ship a monolingual or generation card type; both are template choices you make on a standard note. The official manual covers card templates and review behavior at docs.ankiweb.net/getting-started.html, and community-built Japanese decks worth studying for prompt design are listed at ankiweb.net/shared/decks.