I'm referring to the changes Khatzumoto made to the way he did his (mandarin/cantonese) cards, going from phonetic reading to Hanzi, and applied to Japanese as outlined in his post: http://www.alljapaneseallthetime.com/blo...panese-too

I've just started putting sentences into Anki, I'm about 50 in. Following his advice I'm putting the question in kana, and the answer is the sentence with all Kanji (and English definition until I can go monolingual).

I haven't seen this discussed anywhere else on the forum yet; has anyone had any experience or good results doing it this way (as opposed to Kanji to reading)? Does anyone like/dislike doing it this way? Is there anyone who has tried both methods and is directly able to compare?

Interested in hearing any feedback. It's too soon to tell whether it's a good choice for me yet.

First, production cards take a MUCH more time to review than recognition cards. For me, to do about 10 production cards, I could probably make it through 30-40 recognition cards in that same amount of time. So if you are doing only production cards, you simply wont have a lot of time to do a wide variety of cards.

The only advantage of production cards, as I see it, is that you learn to write the kanji. By doing this, you can start to phase out your Heisig reviews. So I think it makes sense to do SOME production cards to make sure you stay fresh on being able to actually write kanji words, but I would mostly just stick with recognition, because the main hurdle is just learning lots and lots of vocab, so you will want to have lots of cards.

I'm wondering the same thing myself, actually.

I have a plan (but this plan was just implemented a little bit ago, so I have no idea if it's a good idea or not). Anyway, perhaps I'll put most of the sentences in the SRS with the question in kanji and the answer in kana. As Zarxrax notes, I'll be able to get through a lot more using this recognition scheme. But there are some words that I really wanna learn the kanji for. Maybe I'll learn them in context, but my idea is to add a few to the SRS with question=kana and answer=kanji. That way, I can work on production, but it's just one word at a time... No need to write out whole sentences when a sentence might just have one word with kanji I don't yet know.

I know that for meaning in context, sentences might be better than mere vocabulary words... With my idea, I'd still learn meanings through sentences, but for getting a bit of extra production practice, there's be kana-to-kanji vocabulary words sprinkled in the repetitions.

Personally, I think the hardest part is to get the reading of the kanji right so I don't think it would be worth it to go kana -> kanji for me... that is the easiest part yet takes the most time to review, seems ineffective to me.

Besides, you will spend a LOT more time reading kanji than writing them so it seems natural to focus on that.

deebo Wrote:I'm referring to the changes Khatzumoto made to the way he did his (mandarin/cantonese) cards, going from phonetic reading to Hanzi, and applied to Japanese as outlined in his post: http://www.alljapaneseallthetime.com/blo...panese-too

I've just started putting sentences into Anki, I'm about 50 in. Following his advice I'm putting the question in kana, and the answer is the sentence with all Kanji (and English definition until I can go monolingual).

I haven't seen this discussed anywhere else on the forum yet; has anyone had any experience or good results doing it this way (as opposed to Kanji to reading)? Does anyone like/dislike doing it this way? Is there anyone who has tried both methods and is directly able to compare?

Interested in hearing any feedback. It's too soon to tell whether it's a good choice for me yet.

I think this should only be done after a fair amount of Kanji -> Kana cards have been matured. It's just plain slow to do if you are just starting out. I know because I tried.

Then there is that ache you get in your brain as you stare at かい and try to figure out which of the thousands of kanji you have in your brain reprensent it.

This is relatively simple when you have a large Japanese vocabulary, though.

Murjab Wrote:my idea is to add a few to the SRS with question=kana and answer=kanji. That way, I can work on production, but it's just one word at a time...

That might not work out very well, because Japanese has LOTS of homonyms, so there are potentially lots of different words which all have the same kana reading. I guess maybe if you put an english meaning alongside them, or something like that, it could work.

An idea that I have been thinking of trying is to to have a regular recognition sentence with one word written in kana and highlighted. This way you can get full context if you need it, but also just focus on writing the single word.

Zarxrax Wrote:That might not work out very well, because Japanese has LOTS of homonyms, so there are potentially lots of different words which all have the same kana reading. I guess maybe if you put an english meaning alongside them, or something like that, it could work.

Very true... So far I've only put in a few words that don't (as far as I know) have homonyms. But you're right; this could possibly get very confusing... Perhaps sentences are really the way to go for production... Within a sentence, the meaning is (or should be) pretty clear for any homonym... Perhaps The question could include an example sentence with the target vocabulary word highlighted, and the answer, for production purposes, would simple be writing the kanji for that word (and while you're at it, comprehending the whole sentence as a bonus). That way, at least, you don't have to take a long time to write out every word in the question sentence.

I don't really agree with that last post. I agree that it's important to keep sentences short, that makes it easier and faster to review and focuses on the things you actually want to learn. BUT, it also greatly limits the context and the... bonus learning.

I put in this sentence today in my own SRS:
経営の安泰を図る
This is a pretty massive sentence for me since I didn't know ANY of the words when I entered it. Still, it's only 8 signs long. It proves that in just a very short sentence, you can get a LOT of bang for your buck. On the other hand, this just teaches me 3 words, nothing else. The grammar is more or less nonexistant.

IF you want to learn grammar from your sentences (a lot of people here do) longer sentences will be important. He's right that long sentences are mostly just short sentences strung together, but this stringing together of sentences usually hold meaning which will be lost on someone who just know the shorter ones.

I would make an analogy to grammar overall: Knowing all the words in a sentence won't help you if you don't know the grammar.

Summary: Keep it as short as possible but when you feel a sentence is important the way it is because how the grammar works in it, keep it. Don't put a limit on characters just because it's smart to keep sentences short, keep it only as a guide.

nest0r Wrote:What I usually do for those short active recall cards is to add context in the Question section, be it in the form of a picture or a slice of the surrounding sentence or both.

I pretty much do the same thing. When I enter a sentence in I just cut out whatever isn't part of what I want to learn. If the rest of the sentence is necessary for understanding, I just put it in my "context" field.

I hate production cards. I can usually write the kanji for a word after I've seen it enough times, so for me the the cost of production cards is way too high. It helps to have the kanji associated with the onyomi, you know...

I do it both ways, with going from audio to kanji as the first step.

I think this roots the information much more deeply in my mind. The trick is finding ways to avoid it taking so long. Generally I don't write out the whole sentence, but just the part that I am studying. However, this can create the problem of not listening to the meaning of the sentence, rather just trying to find the word (or grammar) that is the focus of the card. So I do all sorts of tricks to stop this, such as repeating after the audio, imagining a situation in which this sentence would be said and so on.

I think the problem with just reading is that the word is less likely to come out when you speak. You can get through many cards by just knowing the readings of the kanji, without actually remembering the word at all. This will generally resolve itself though because you will encounter the word in the wild and it will eventually flow out naturally when you speak. The problem is that eventually you are left with words that you will rarely encounter so it is hard to get them into a usable state this way.

Plus as a rule with language learning you want to attack it from as many angles as possible. Our brains are just too deft at avoiding remembering anything that we don't stuff down it's throat (if you get me).

This is me anyway, I think everyone learns differently though.

Guess it makes more sense to call them Dictation cards than production cards. Like others, I switched to kana to kanji dictation then to TTS+kana to kanji dictation cards. Like the above said, this takes A LOT of time, especially if you're doing the entire sentences.

I think Tobberoth has it right that grammar cards at times need to be lengthy if the purpose of the card is grammar. That said, there's little need to do dictation on grammar cards. Maybe do audio recognition and reading recognition. But to waste 2 to 3 minutes writing out a sentence that's just showing how "~kara" is used (sorry, don't have an IME on this compute).

So, just don't treat all cards equally. If it's a grammar card, don't dictate it. If it's a vocabulary card (you added it for a new word), perhaps bold term the new word portion of the sentence and dictate just that word and not the whole sentence. This can help speed up your reviews, especially as you build up sentences.

I find dictation has very good merit. If you hear "kouka", see a photo of over the counter medicine and then hear the sentence "kono kusuri wa i o mamoru kouka ga iru." With that, if you know the meaning and can write the correct kanji for kouka, you probably have the pronunciation down (gods this is annoying without an IME, sorry again guys). Plus, as you did it all with audio, your listening probably is not too bad either.

Reading (well, comprehension) cards also have merit. With context provided in dictation, you can sometimes power guess the kanji. However, with reading, you have to know the on-yomi and kun-yomi in addition to the meaning. So I say don't just do dictation or reading. Do both, and the easier ones will advance faster and get reviewed less thanks to the miracles of the SRS.

thermal Wrote:I think the problem with just reading is that the word is less likely to come out when you speak. You can get through many cards by just knowing the readings of the kanji, without actually remembering the word at all. This will generally resolve itself though because you will encounter the word in the wild and it will eventually flow out naturally when you speak. The problem is that eventually you are left with words that you will rarely encounter so it is hard to get them into a usable state this way.

This is just me, not what everybody should do, but I don't think the SRS has to teach me everything. I prefer learning how to actively recall words like you say, encountering the word in the wild enough. Really, though, if a word is that rare, shouldn't it be as rarely used by us?

Zarxrax Wrote:

Murjab Wrote:my idea is to add a few to the SRS with question=kana and answer=kanji. That way, I can work on production, but it's just one word at a time...

That might not work out very well, because Japanese has LOTS of homonyms, so there are potentially lots of different words which all have the same kana reading. I guess maybe if you put an english meaning alongside them, or something like that, it could work.

I don't think Japanese homonyms is an excuse for not practicing kana to kanji. After all, when you listen to Japanese, you don't get the benefit of kanji, do you? Practicing kana to kanji is a way of practicing "listening" but at your own pace, as a precursor to actual listening practice.

Anyway, I've started putting sentences from podcasts into my SRS as kana to kanji. It's only early days yet, but it seems to be working OK so far.

plus, you can always do the "image hack" that Khatz was talking about for kana to kanji cards. That might help the 11-bujilllion homonyms...