Spaced repetition is probably the reason you can recognize as many kanji as you can. Apps like Anki and WaniKani schedule each item to come back right around the moment you are about to forget it, which is the most efficient way to move something into long-term memory. It works. The problem is what most of us point it at.
The task is the thing
A spaced repetition system does not care what the review task is. It only schedules. If your review task is “look at this card and judge whether you knew it,” you are training recognition. If your review task is “write this character from memory,” you are training recall. Same engine, completely different result.
This is the lever almost no one pulls. We run thousands of recognition reviews and zero writing reviews, then wonder why we can read but not write. The method was never the issue. The task was.
What a writing review looks like
A good writing review is simple:
- You are shown a meaning and reading, not the character.
- You write the character by hand, from memory.
- You grade honestly: could you produce it, or not? Stroke-order slips count.
- The schedule uses that grade. Missed characters return soon; clean ones return later.
That honest self-grade is the whole game. Recognition lets you fool yourself (“I would have known that”). Writing does not. The page is either right or it is not, which makes the signal feeding the scheduler far more accurate. For why that distinction matters so much, see recognition versus recall.
How your grade sets the interval
The honest self-grade is the whole engine. It decides when each character comes back:
| Your honest grade | What it means | When it returns |
|---|---|---|
| Could not write it | Recall failed | Very soon |
| Wrote it with effort or a peek | Shaky | A few days |
| Wrote it cleanly from memory | Solid | A longer interval, then longer |
Because writing gives a clear right-or-wrong signal, the schedule it feeds is far more accurate than one fed by “I would have known that.”
Why it stays sustainable
The fear is that adding writing doubles your workload. It does not, for two reasons.
- Spacing keeps volume low. You only review what is due, and well-known characters drift to long intervals. Most days are a handful of items, not hundreds.
- Writing reps are worth more. One act of recall builds more durable memory than several recognition reps, so you need fewer of them per character.
A few minutes a day, every day, is enough to keep your whole set writable. The daily habit matters more than the session length, the same way making stroke-order practice a daily loop is what turns it into muscle memory.
A realistic daily session
A writing SRS looks unremarkable on any given day, and that is the point. You open it, it shows the dozen or so characters that are due, and for each you see a meaning and reading and write the character by hand before checking. A clean one takes a few seconds; a missed one you rewrite once correctly, and it returns sooner. Ten to fifteen items, five to ten minutes, and you are done. The volume stays low precisely because the schedule hides everything that is not due yet, and because one writing rep is worth several recognition taps. There is no marathon session and no streak to protect, just a few honest minutes that keep the whole set writable.
Putting it together
Recognition study and writing study are complementary. Keep your reading SRS for meaning and vocabulary. Add a writing SRS for production. Together they cover the whole skill instead of half of it.
Kanji Write Practice is the writing half, built as a spaced repetition system for handwritten recall: it prompts you with meaning and reading, gives you a canvas and a stroke-order guide, and schedules each kanji based on whether you could actually write it. If you want spaced repetition working for your writing and not just your reading, join the early-access waitlist.
