In other words - English on the front and Japanese on the back?

This is my 2円, but it doesn't seem bad to me if you've got a Japanese sentence as context on the front as well.

The problem with going English word → Japanese word is that there will be several possibilities in Japanese to give to the English one. Whereas with Japanese word → English word, anything that makes you understand the Japanese is ok.

With studying Japanese with flashcards, we want to test ourselves on specific Japanese words rather than several possibilities to translate English ones. I think the clozed Japanese sentence on the front would to give a clear enough context that there's no such confusion, even in a deck that eventually has 2000 cards in the reviewing cycle.

That said, if you read up on how to customize Anki decks you'll probably find that it isn't difficult to rearrange the fields on the cards--what appears on the front and on the back--without losing progress or anything.

You're fine. You can keep those facts as they are, or you can flip them. Doesn't matter (well, flipping them would make it a little easier to continue reviewing them). There's absolutely no need to add a second bunch of facts, this time the other way around. That would be a complete waste of time.

But you should stop adding new facts that way. It's much harder and more confusing (because you could answer with a perfectly correct Japanese sentence and fail the card, since there is more than one way to say something in Japanese). It's better to just go from Kanji to reading/meaning, or from reading to Kanji/meaning (or even both, in two different facts, I think that's all a matter of personal preference). If you're going reading-> Kanji/meaning, the question should be the audio, not the hiragana (for a few reasons).

It's not "bad", but recognition is deemed to be quicker (and lets you move on to reading).

I find mixing the two can be beneficial (iknow style) for learning a new word; then use recognition for long term memory.

Alternatively you can do it like this (using s0apguns example) if you want to train listening comprehension and active kanji recall:

Front:

(sentence audio plays)

かくち

[write the word in question in your mind's eye, but don't worry about the rest, just get the meaning of the sentence]
Back:

Picture

各地で大雨が降っています。

every place, various parts ー 各地[かくち]

(button to repeat sentence audio)

各地[かくち]で大雨[おおあめ]が降[ふ]っています。

Heavy rain is falling in all regions.

***[hiragana in brackets show as furigana above the kanji***

Takes longer than passive review and is sort of what mmhorii is emphasizing without doing English--> Japanese

Stansfield123 wrote:

There's absolutely no need to add a second bunch of facts, this time the other way around. That would be a complete waste of time.

In my deck I have forward and reverse cards for each of my notes (i.e. Japanese -> English and English -> Japanese). Is this what you mean? It does take a bit longer, but I feel like just doing Japanese -> English is too easy and would make it harder to recall words when attempting to form sentences. Has anyone tried both ways and have any commentary on how each worked for them?

Any links to studies done on this kind of thing?

EDIT: After thinking about it for a bit, I think I might try out recall-only studying. I don't think it will save me a lot of time (the recognition cards are always super quick, anyway), but I might as well save time where I can.

Last edited by Acedio (2013 March 06, 6:53 pm)

Acedio wrote:

In my deck I have forward and reverse cards for each of my notes (i.e. Japanese -> English and English -> Japanese). Is this what you mean? It does take a bit longer, but I feel like just doing Japanese -> English is too easy and would make it harder to recall words when attempting to form sentences. Has anyone tried both ways and have any commentary on how each worked for them?

Any links to studies done on this kind of thing?

EDIT: After thinking about it for a bit, I think I might try out recall-only studying. I don't think it will save me a lot of time (the recognition cards are always super quick, anyway), but I might as well save time where I can.

What I mean is that I don't recommend English -> Japanese cards, but, if you do them, then also doing the same fact in a Japanese -> English  card is pretty pointless. I'm not asking you to trust me, I have a logical explanation why. Here's my reasoning:

The only splitting up of a fact into two cards that makes sense to me, in Core2K, is Kanji -> Reading and then Audio -> Kanji. I like that because each fact has two cards, but you produce two different things with each (two-carded) fact: Kanji from one, reading from the other, and you also have two different forms of input: reading and listening. That is probably worth the fact that you're doing significantly more reviews.

However, with English -> Japanese, you produce two different things with just this one card: you produce both Kanji and reading (NOTE↓). That's already bad, because it goes against the SRS principle of making answers simple  (comprised of one simple piece of information). When you also add a Kanji -> Reading card, you produce a reading again. But you already produced a reading (which is the easiest part of it anyway), so you're wasting time by doubling up on the easiest part of your studies. It makes no sense to me.

In conclusion: doubling up on on each fact, by doing it twice: Kanji -> Reading and Audio -> Kanji covers everything: both types of production, both types of input, and listening comprehension is the cherry on top of this beautiful cake.  If completeness is what you're going for, this is the way to go.

The only two alternatives to this that make sense to me are:
1. only doing one or the other (you can even switch periodically, to break up the monotony). This only covers one type of production and one type of input, but saves about 25% to 33% of your time (meaning you get to go through that much more material, but less thoroughly). Like I said, it's a matter of personal preference whether you want to be thorough, or you want volume. Imo, both work fine.

2. If you're doing Anki reviews in conjunction with heavy input (reading and listening to native material), then producing Kanji writings of single words/fact is all you need, as far as Anki is concerned (the way I described it in some of my other recent posts in various threads, by setting the answer as the writing of a word, and the question as a sentence you can easily read, that contains that word - but the word itself is taken out of this question sentence, replaced with its reading, or even a blank for advanced students, instead).

This second alternative is what I've switched to recently. But, for this to work, reading and other forms of input are an important component. Otherwise, you're way too focused on just one aspect of the language (the writing system), at the expense of everything else. It also takes a few minutes/ fact to find the example sentence and add the fact to Anki.

Note: my logic assumes that trying to produce Kanji without first establishing the reading of a word is stupid. I'm hoping this is a self evident statement, since the last thing this post needs is another paragraph to back that up. I've been droning on long enough.

Note2: comparing Japanese sentence reviews to let's say French sentence reviews is a mistake. With Japanese sentence reviews, Japanese to English cards do in fact force you to produce Japanese (either the Kanji, based on Japanese audio, or the reading, based on the Kanji). A French -> English sentence would be pure input, and therefor the usual arguments concerning the pointlessness of drilling input would apply.

Note3: Oups, in Anki 2 facts are actually called notes. So, for the purposes of this post, fact=note. I had a nagging feeling that I'm getting something wrong, through the typing of this whole post.

Last edited by Stansfield123 (2013 March 08, 2:13 pm)

Thanks for the detailed reply, Stansfield

Stansfield123 wrote:

However, with English -> Japanese, you produce two different things with just this one card: you produce both Kanji and reading (NOTE↓). That's already bad, because it goes against the SRS principle of making answers simple  (comprised of one simple piece of information).

If your cards go from Word (in Kanji) -> Reading and Audio -> Writing (in Kanji), isn't meaning implied as an output for both of them as well? If you blank on a meaning, but can produce the reading and/or kanji, do you mark the card as failed?

Stansfield123 wrote:

2. If you're doing Anki reviews in conjunction with heavy input (reading and listening to native material), then producing Kanji writings of single words/fact is all you need, as far as Anki is concerned (the way I described it in some of my other recent posts in various threads, by setting the answer as the writing of a word, and the question as a sentence you can easily read, that contains that word - but the word itself is taken out of this question sentence, replaced with its reading, or even a blank for advanced students, instead).

I like this approach. You could go either Sentence + Reading -> Kanji + Word Meaning (as you  explained above) or Sentence + Word Meaning -> Kanji + Reading (e.g. "(That)はかわいい犬です。"). As you mentioned, reading and listening would be very important with this method.