RECENT TOPICS » View all
An endless discussion regarding what's the best setup (production, recognition etc.) for your cards in SRS.
I used English on question side and then kanji, kana and audio on answer side before I today thought that maybe using audio on front would be better since my brain maybe wants to think English→日本語 later on if I stick with my current setup.
Also wouldn't this also benefit my listening comprehension?
So what do you think? Should I change to audio instead?
Last edited by daevil (2013 June 21, 3:45 pm)
i think there are some drawbacks, but of course some benefits.
1. japanese audio -> japanese text is more recognition than producing. you dont have to produce the reading.
2. readings aren't enough to know a word. there are words with same reading, but different meaning. so you need context for this. (but same problem for english->japanese)
i would mix things up. (or more precisely I mix things up in my anki setup).
one note can have many cards. some audio-> japanese, audio-> english, englih-> japanese, japanese-> english, etc.
We were just talking about this in this thread.
http://forum.koohii.com/viewtopic.php?id=11423
I still stick by what I said there, that none of the methods are perfect and you might as well just go with what motivates you.
I see no problem with doing audio cards, as long as your audio is generic enough that you don't just start to recognize the voice rather than the word. Or you could just put the kana up on the front. The problem is, you're going to have a ton of homophones, so you need a way to deal with that. It happens pretty often that 5-10 common words will have the same reading. You could study all the homophones as a group, but that will make for difficult cards, especially if you want to test writing as well. Another alternative is putting every homophone (except the word for that specific card) on the front. That means by process of elimination, you can come up with a single answer.
Another thing- if you're going to test writing, I'd advise not physically writing it out but just recalling the keywords. Leave keyword->writing testing to your RTK cards.
Haych wrote:
We were just talking about this in this thread.
http://forum.koohii.com/viewtopic.php?id=11423
I still stick by what I said there, that none of the methods are perfect and you might as well just go with what motivates you.
I see no problem with doing audio cards, as long as your audio is generic enough that you don't just start to recognize the voice rather than the word. Or you could just put the kana up on the front. The problem is, you're going to have a ton of homophones, so you need a way to deal with that. It happens pretty often that 5-10 common words will have the same reading. You could study all the homophones as a group, but that will make for difficult cards, especially if you want to test writing as well. Another alternative is putting every homophone (except the word for that specific card) on the front. That means by process of elimination, you can come up with a single answer.
Another thing- if you're going to test writing, I'd advise not physically writing it out but just recalling the keywords. Leave keyword->writing testing to your RTK cards.
Okay, but why shouldn't I write it down? Is RTK alone enough?
Also, I read your post in that link (interesting read btw) and by "meaning" do you mean the English keyword/definition by that?
So maybe it would be better with 2 cards for every vocab? One with production and one with recognition? Also if so do you recommend me not to write the answers on any of them?
--
Not writing anything down on my vocab decks (grammar also then) would REALLY make me do my reps faster and give me more time on something that some people call "real Japanese". Like read Japanese texts and such instead of brute forcing in information into my head.
Treating Anki like a tool and not my main learning source would maybe benefit me more.
Anyone who has made this change and can tell me if it resulted in something more effective?
Last edited by daevil (2013 June 21, 6:08 pm)
Not writing it down is just a tip to save you time. The act of writing will probably slow the average card to around 30 seconds. Without writing, you can easily get answer times less than 10 seconds average. Three times slower really starts to add up.
In that post, by 'meaning' I essentially just meant an understanding of how the vocab word is used. You'll probably use an english definition, but the point is, you should keep the language-independent sense of 'meaning' in mind.
Also, your use of 'keywords' to refer to the meaning of vocabulary disturbs me a bit. Are you using 1-word definitions for vocab words? If so, stop. I know some insist on keeping SRS cards minimalist, but that is just too much. You need to keep your definitions vivid if you're going to have any chance to remember them. One word is highly inadequate, to say the least.
And now you are considering production and recognition rather than audio? It's not such a bad idea. Production will still be as difficult as usual, but recognition will serve to reinforce it. Keep in mind you'll still have the problems that I had with ambiguity, and you should read my 2nd post for that. Recognition is the only method that involves no ambiguity right off the bat. Also keep in mind that doing both will slow you down a little. If you can't learn 30 new words a day doing this, I'd say get rid of one card type.
Yep, writing do takes me like 30 sec each cards.
Regarding the 'keywords', this is what I am using currently:
FRONT:
Tuesday
BACK:
火曜日
かようび[sound:かようび - 火曜日.mp3]
But then keeping those cards and adding (on new cards) an additional card using this:
FRONT:
火曜日
BACK:
かようび[sound:かようび - 火曜日.mp3]
Tuesday
...would probably serve well. Or what do you think?
Err, yeah. The first is production and the second is recognition. If you want to add two cards for that example, go ahead, I guess. But I think for simple ones like that, just one card is fine. The benefits are more for the case where you have multiple meanings like:
務める
つとめる
1: to be employed at (some employer);
2: to fill the job position of _______;
3: to conduct a religious service
Production would ensure you can think of the word while speaking or writing, but it wouldn't ensure that you can remember the 3 meanings exactly. Since its more complicated, your memory of the definition might be a little vague. That's where recognition comes in. That gives you a chance to test your knowledge of the definitions and solidify it a bit better.
Haha, I maybe gave you a bad example. But still dual cards on those words will give me double exposure and they are usually very easy to complete.
Well, I'll think I will test this out and see how it goes.
Thanks.
I typically do the standard Kanji -> Kana/English cards. At the moment I am experimenting with some production cards, but only for cards marked for the JLPT. I am doing these with higher (200%) intervals as well, as I have already covered these words in recognition.
My production cards are like so:
I think listening comprehension is mostly limited by grammar and not vocabulary.
It's no good knowing all of the vocabulary in a sentence when you've yet to understand how the words are supposed to relate to each other. On the other hand, knowing where to look out for grammatical constructs helps you figure out what someone is saying, even if you don't know some or even any of the vocab they're using.
The key ingredient that facilitates long-term storage is meaningfulness. This term refers not to the inherent interest or worthiness of information, but rather to the degree to which it can be related to information already stored in our long-term memory. One concept or piece of information is more meaningful than another if the learner can make a larger number of connections between that piece of information and other information already in long-term memory.
http://education.purduecal.edu/Vockell/ … 6_long.htm
In other words, memorising new words is a lot easier if you've already learnt an abundance of information relevant to how words can be used. If you try to memorise vocabulary without grammar, too many hours are wasted on "precomprehension" (only being able to recall them via mnemonics or rote memorisation) -- hours better spent on grammar.
I personally use the kanji (front), kana/audio/meaning(s) (back) setup to learn readings, but it doubt I'll be going beyond core2k until I've finished hugo japanese & the dictionaries of basic and intermediate japanese grammar. I'd rather "work my way up" than wait for more fragmented knowledge to piece together and consolidate what I've learnt. It's not as simple as this, but I think I make a decent point.
Also agree with Haych about skipping the writing on vocabulary cards. It's just too much time spent per card.
Animosophy wrote:
Also agree with Haych about skipping the writing on vocabulary cards. It's just too much time spent per card.
Tried doing without writing and it felt wrong. My thoughts when I was trying to recall the kanji was thinking about the English keyword as well as the kanji. But when I write I only think of the kanji.
Don't you also experience this?
Last edited by daevil (2013 June 24, 5:48 pm)
daevil wrote:
Tried doing without writing and it felt wrong. My thoughts when I was trying to recall the kanji was thinking about the English keyword as well as the kanji. But when I write I only think of the kanji.
Don't you also experience this?
Honestly I don't know what to advise. I'm just another learner and I've frequently changed the way I study since the beginning (well, post-RTK at least).
My opinion on vocabulary cards is that they're not worth the time required to write answers down. Even if it helps, it's a major net loss. If I did the same for my core2k recognition cards, I'd cover a lot less words for a slightly better retention stat. No writing is the only way I can comfortably do 50 new cards a day and keep up with reviews.
I still think writing Japanese is important though, so I write example sentences for grammar. It's better for handwriting because they're not just individual words, and it reinforces the grammar. S'all I got.

