I added a new card type to my review system a little while ago, and it seems to be working out OK. Here is a writeup in case anyone is interested. My deck has 4 fields, listed below.

ja — Japanese sentence written using kanji

kana — sentence spelled in kana

en — English translation

audio — audio recording of sentences

I now have 4 types of cards, with the Writing type being the new one:

Listening: hear sentence, understand meaning and type kana

Kanji: hear sentence and see sentence with 1 kanji missing, handwrite missing kanji

Reading: see sentence, understand meaning and type kana

Writing: hear sentence, understand meaning and write sentence by hand

In theory, if you can do Listening and Kanji, you should be able to do Writing, but I have found it doesn’t work out that way, for several reasons. These reasons are why I think the extra card type is useful.

1. I didn’t know the kana as well as I thought. There were some sentences which I failed only because I forgot what hiragana “e” looks like, or katakana “u.” Probably if you asked me to sound out a word that used them, I wouldn’t have a problem. And yet I failed anyway. I found I was starting to rely too much on context to recognize kana, and wasn’t really very strong on them individually. The writing practice makes you pay attention and get it right, which in turn makes recognition faster.

2. My listening comprehension wasn’t as good as I thought. Part of this I noticed when I stopped allowing myself to hear the audio more than once on the listening cards, so this isn’t really about the writing per se, but the writing cards make the problem more obvious.

When you have only one chance to hear a sentence and then have to type it out in kana or handwrite it, it puts a strain on your short-term memory. If there is even one word you struggle to understand, your brain will stall while trying to remember what it means, and while that is happening, it has to also try to process the rest of the sentence it is hearing. So you end up playing this juggling act where you both try to simultaneously hold on to the fleeting memory of what you just heard, and dredge up some weak associations you have for a word. Usually you can’t do both at the same time, unless it’s a really short sentence, so you fail.

That is, unless your comprehension is flawless. If you know the instant you hear the sentence what everything means, and how it all fits together, and you have no doubt over what particles go where (に and で can be tricky, etc.), you can just unspool your memory of what you just heard into your hands and be done with it. But if you have to stop and think about one part of a sentence, even for a second, it is likely you will forget some other part of the sentence and fail.

The writing practice makes this problem even worse, because in addition to remembering the sentence itself, you have to remember all the strokes of every kanji in the sentence, and if you can’t to that effortlessly you are likely to fail. So the writing practice sort of becomes the ultimate test of whether you have really mastered listening comprehension and kanji. Obviously you need a large enough deck to prevent yourself from just memorizing each sentence, but if you look around for Core10k, that shouldn’t be a problem.

3. My kanji knowledge wasn’t as good as I thought. Most of the problem is in how the individual kanji cards in my deck work. Basically, the problem is that they don’t test every way each kanji is used. I haven’t figured out a good way to do that yet, so even if a given kanji card becomes mature, it doesn’t mean I have studied all possible ways that kanji is used in the deck. It isn’t feasible to create a kanji card for each occurrence of each kanji in each sentence, because that would require adding maybe 5 additional cards per sentence, on average, which is a massive number of cards, many of which would be equivalent. If you know 少し in one sentence, that’s enough. You don’t need dozens of cards just for that.

So unless you can figure out a good, quick way of identifying unique types of usages (which you might be able to do using Mecab, but I don’t know for sure), you have to take a scattershot approach. So weaknesses have slipped in to my understanding, and for example I end up with the vague idea that ぶ can be written as 部 or 分, and those kanji can be used with roughly the same meaning, but when it comes time to write だいぶ I struggle. I have similar problems with 始 and 初, and 合 and 会. Writing each sentence by hand helps isolate these kinds of problems so I can fix them, while limiting the review burden to one additional card per sentence.

There are of course some problems. One big one is that there is often more than one way valid way to write a sentence. So you have to get to know what the variations are, so you know what to accept as a pass. That can lead to some uncertainty about what is really correct. Another one is that writing everything out by hand takes time, especially at the beginning when your skills are weak. But you start to see improvement in your speed in just a few weeks, so there is a sense of accomplishment that helps justify the time spent.

Overall I think adding more writing practice has been useful, hence this writeup. I have heard more than one person say it is not worth the time, since handwriting isn’t a very useful skill in the age of computers. I don’t want to spend too much time arguing about it, because people can think whatever they want. But I will say that even if the skill isn’t useful per se, doing writing practice can help you solidify your other skills, as I have explained above. I will also say that adding a new card type that is different from the other card types makes your reviews more interesting, because of the added variety. And if you are talking about a large deck that is going to take a long time to finish, you are going to need all the help you can get when it comes to staying motivated. Just make sure to use larger intervals to compensate for the fact you are seeing each sentence on multiple cards.