I've been wanting to give something back to this community for a while, and here it is.

If you've been using Anki to study vocabulary, you may want to stop until you take a look at this. It's a shared deck called 2-Step Japanese.

It's basically an Anki hack, which makes a radical improvement (I'd say) to the program.

I’d been wanting to do this for years, and about 6 months ago I finally completed it, and have been testing it ever since. It’s a bigger improvement than I ever imagined. I've been using it every day, and you can probably tell I'm excited about it, so let me explain:

In normal Anki, you only have 1 step. So your cards either look something like this:

Front: 鉛筆 　 Back: 　えんぴつ　Pencil

or like this:

Front: これは鉛筆ですか？ 　Back: えんぴつ Is this a pencil?

That is, you’ve either only got vocabulary with no context, or a sentence, which is good for context, but makes it too easy to guess the kanji.

Enter 2-Step Japanese

This method solves that issue. The cards are presented differently than normal. Like this:

Step 1: The kanji is shown. 鉛筆

Then, when you click on the kanji,

Step 2: The sentence is shown. これは鉛筆ですか？

At this point, you’re back to regular Anki. Either click “Show Answer” or the spacebar, and the kanji with furigana will appear, along with an English definition. (You’ll need to install the “Japanese Support” add-on to see the furigana.)

Just reading about this, it may be hard to imagine what a difference this makes. Download it and try it! I think you’ll like it. I wouldn't study vocabulary any other way. But then I came up with it, so I'm kind of partial.

This is a sample deck with only 288 cards at roughly the JLPT2 level, so feel free to delete the content and add your own material. The content is far less important than the 2-step method.

You’ll note that when editing the cards, the fields are not in the order they are shown when reviewing the cards. This enables you to use multiple example sentences for the same kanji without having to turn off the duplicates checking. It also makes it easier to paste in sample sentences. It feels weird at first, but once you get used to it, well, maybe it’ll make sense. You could always re-order them if you want, but try it this way for a while and see.

I’ll also eventually be uploading a 3000+ card deck using this 2-step method after I get that cleaned up.

Let me know what you think.

Enjoy!

At the risk of sounding rude, please allow me to try and simplify this to see if I understand the process correct:

>vocabulary term
>test vocabulary term (in some manner- writing it down, etc.)
>click on vocabulary term to be provided with a related context
>use context to see how accurate your prior understanding of the vocabulary term was


Is this basically the process? If so, I like it. Like it a lot, actually.

EDIT:
Based on my assumption, it would be really, really useful if step 2 could have a randomized pool of related sentences (5, 10, 15 sentences, whatever) to provide a dynamic context over the life of the step 1 vocabulary term, as opposed to a static context with a single related sentence. It could provide a more in-depth understanding of the meaning and usage of the vocabulary item in Step 1.

Though I have no idea how one would make this possible, having a single card/Step 1 with multiple facts for Step 2, without messing up the scheduling algorithm. But if this were possible, I would be happy to purchase such a product -or in some way provide a bit of a monetary "thank you" for the people who made it.

Why not just have two different cards for each fact, one that tests kanji->reading with just the word on the front side and put the sentence on the back side, and one that tests sentence->meaning with the whole sentence on the front side?

My deck doesn't do that exactly, I put sentence on the front side in every case, but test 'sentence with kanji word'->(typed in) reading and 'sentence with kana word'->meaning.

But anyway, the ability to create multiple cards from one fact is one of anki's power, allowing you to test multiple facets of your knowledge.

uisukii Wrote:At the risk of sounding rude, please allow me to try and simplify this to see if I understand the process correct:

>vocabulary term
>test vocabulary term (in some manner- writing it down, etc.)
>click on vocabulary term to be provided with a related context
>use context to see how accurate your prior understanding of the vocabulary term was

Is this basically the process? If so, I like it. Like it a lot, actually.

This is basically the process, though the "test" I use is basically recognition, rather than writing it down. 2-Step Japanese tests you twice for each vocabulary word. Once without context, and once with context.

SomeCallMeChris Wrote:Why not just have two different cards for each fact, one that tests kanji->reading with just the word on the front side and put the sentence on the back side, and one that tests sentence->meaning with the whole sentence on the front side?

Sorry, but that doesn't accomplish the same thing. Both of those setups have drawbacks.

Kanji->Reading lacks context, which makes it harder to remember facts. By creating an example sentence, you provide yourself with a meaningful way in which the word can be used.

Sentence->Meaning is too easy. Using sentences is too much of a crutch, because you're too likely to remember the kanji rather than actually forcing yourself to read it. I'm sure anyone who has used a sentence-based approach has encountered this problem.

The 2-Step method solves both of those issues. It ensures you can read the word without any context, then gives you the context that is so important for memorization.

Do you have any idea on the possibility of allowing for a selection of related sentences in Step 2, without having to create multiple cards for Step 1? Something which would allow for Step 1 vocabulary fact to be, for instance, graded and the next time it shows up it may have any other of the allocated sentences related to the vocabulary term, without altering the scheduling Anki uses for the fact/vocabulary term in Step 1?

As someone with no experience or education in programming, etc. I have no idea if such a concept would be feasible or worth the effort.

uisukii Wrote:Do you have any idea on the possibility of allowing for a selection of related sentences in Step 2, without having to create multiple cards for Step 1? Something which would allow for Step 1 vocabulary fact to be, for instance, graded and the next time it shows up it may have any other of the allocated sentences related to the vocabulary term, without altering the scheduling Anki uses for the fact/vocabulary term in Step 1?

I had thought of that exact thing. I don't think it would be easy to do however. I had considered contracting with a programmer to accomplish that. But then I came up with this method, which if not perfect, is a significant advancement. It makes studying vocabulary much more effective.

Hmmm, that's a reasonable answer. I'm not sure I agree with it entirely, as I find that the 'sentence problem' is self-solving. Once you have enough sentences, you can't instantly recognize them from the first word or two and know the answer before you reach it (although that problem can push a vocab word out quite a bit before you finally fail it because you were recognizing the sentence, not the word; but at that point, you've self-corrected so.)

Uisikii's question I think strikes at the heart of the weakness of this method though, namely that quizzing a word alone as in step-1 leaves you in many cases having to think up many definitions or else be only 'partly' right (at which point how do you grade the card? fail or pass if you missed the 5th definition out of 7 definitions?)

Personally, if I take up the two step method (which I'm vaguely considering), I'd go ahead and have separate cards, and first step to second step would be essentially a pure reading question, or reading and 'general' meaning, and context sentence a specific-meaning quiz. But I'd want to type in the answer for the reading quiz (otherwise it's easy to fool yourself with long-vowel vs. short vowel and did I -really- remember the っ-breath stop sorts of pronunciation problems that are hard for foreign learners of Japanese). It doesn't seem like I can type-quiz the answer and still have two stages, but I haven't installed the deck yet. I think I'm intrigued enough to peek at it tomorrow though.

Don't take my criticism personally or anything please, I just push on these things to get clarity on what the benefits and drawbacks -really- are, especially when things are touted as a new and better way to study.

SomeCallMeChris Wrote:Uisikii's question I think strikes at the heart of the weakness of this method though, namely that quizzing a word alone as in step-1 leaves you in many cases having to think up many definitions or else be only 'partly' right (at which point how do you grade the card? fail or pass if you missed the 5th definition out of 7 definitions?)

Just to clarify: the 2-Step method isn't for stand-alone kanji, but for actual words. You're right, a kanji can have many definitions. But actual words are pretty specific. It still presents something of a "how do I score this?" issue, but after several months of using it, I haven't found that to be a big deal.

Try this! I think you'll see in about five minutes how nice it is.

I understand that you mean words.... have you really -looked- at the definition for 掛ける?
http://dic.yahoo.co.jp/detail?p=%E6%8E%9...=0&dtype=3

Worst-case scenario, granted (and even worse if you only spell it phonetically and don't limit it by kanji.) Still, there are plenty of words with multiple definitions.

Downloaded, now time to upgrade Anki 1 again. Looking forward to playing around with the innards of your hard work.

Edit:
Okay... I don't like the new Anki as much as the old one, but I had to upgrade in order to use your deck. I'm not going back to the old Anki now. It took about 20 seconds to realise that this is going to be a really useful format for vocab study, and the way you've designed this honestly seems to provide a memory reference for the vocabulary word in a way in which static sentences do not. A little like how RTK uses mnemonics to help provide a structure for memorising kanji, this "reveal" approach provides context for the term, but without the actual sentence being the focal point of the fact. Now I'm going to have fun with playing around with this.

An "Anki hack" indeed. Deceptively simple. Like 納豆.

Lovely template. Much respect, my dude.

So far I'm liking it, I already have my decks set up with the word being tested in it's own field so I just copied the hack into my current sentence decks. Definitely interesting, I'll keep using it and see how it goes. Thanks, JapaneseRuleOf7!

Thank you so much for sharing this! I'm finding it extremely useful. =)

I like this idea and will try it out with my own vocab. From the quick glance I gave this deck, the advantages I saw to this method seems to be not ignoring the word inside a sentence, which I often do subconsciously. But anyho, time will tell. Thanks for this useful deck template!

I knew you guys would like it! It's been the single biggest boost to my Japanese learning since I started a decade ago. Thanks for trying it!

I wouldn't call this "hack", you are only using potential of web browser used for Cards presentation .

For same reason (sentences are too big spoiler for vocab words) I use similar method for more than half year and it makes wonders for my recognition retention. In my case I use :hover on hidden sentence instead of onclick because it allows me to easily hide sentence again when i leave area with mouse.

So my "Study" flow for Recognition deck is like:
1/ Only vocab word is shown

http://imageshack.us/a/img834/6398/questionk.jpg

2/ On mouse hover sentence is shown

http://imageshack.us/a/img708/6849/questionhover.jpg

3/ Answer is shown

http://imageshack.us/a/img18/1156/answerm.jpg

Fortunately Optimized Core 2k/6k deck has Vocab words inside <b> element in Reading field so there is no need to change anything, only Card template and css.

Btw. If someone is interested, i can post Card template and css. But it contains references to fields present only in my deck (like Czech translation etc.) so it'll not be copy&paste only.

Edit: My card template is based on netsplitter work posted in Nukemarine's sugestion thread, so some credits goes to him.

I think I've been doing something similar for ages now.
Only on my deck I don't have that cumbersome "clicking" step.

I simply have my vocabulary word displayed on the card in a large font, then the context sentence is printed right under it. When I review, I just look at the vocabulary word and try to figure out what it is, and then I am free to read the sentence for more context if I want.

I guess the only difference here, is in my method, the context sentence is already shown, wheras in your method, you have to click in order to get it to appear.
My method is probably a lot easier to set up and use, I think.

And honestly, I have never had any problem with "cheating" and looking at the sentence before I actually intend to. Maybe its just that my japanese level is still low, but I have to put quite a bit of effort into looking at a sentence before I can read it, unlike in english where I could read a sentence with just a quick glance.

I hope I'm missing something here, so please inform me if so. I' using Kore2k/6k, and it's like this:
Front: Vocabulary in Kanji written in big font + sentence in Kanji in small font

Back: vocabulary meaning + sentence with Furigana + sentence meaning

When the next card is shown during reviewing I only look at the word, and answer both reading and meaning. If the Kanji can have different meanings/readings I look at the sentence, otherwise I don't look! (no cheating). And words without context aren't giving me any trouble mainly because I've done RTK, I think. Also, I make sure to differentiate between words with same Kanji clearly like 見せる、見える、見る... to avoid confusion in the future, so I don't look at the sentences for such words too.

So, I don't see how this new method will improve my studying. Please tell me if I'm missing something or if I'm doing something wrong.

Thanks for sharing and trying to help

Love it so far. Turned my vocab deck into 2-Step, the only difference is that I have pictures. Is it worth getting a picture from google image for each word? It helps me quite a bit with answers, but like the sentences it may affect my actual knowledge(I may guess the meaning right thanks to the picture).

@undead_saif
question 1. word
question 2. sentence
answer

It's like answering a question with a question--you get a thorough understanding of the content. The emulation of multiple questions for one answer also creates the following effects: comprehension, clarity, and immediate feedback.

I assume 'looking at the sentence only if you really need to' makes you perceive the information differently.

meeatcookies Wrote:Love it so far. Turned my vocab deck into 2-Step, the only difference is that I have pictures. Is it worth getting a picture from google image for each word? It helps me quite a bit with answers, but like the sentences it may affect my actual knowledge(I may guess the meaning right thanks to the picture).

Yes, I use pictures a lot in this method. There's one image included in the sample deck that shows how I've set this up. You're right that if you include the picture in Step 1, it will give you too much of a clue. You want the picture in Step 2.

Aspiring Wrote:I assume 'looking at the sentence only if you really need to' makes you perceive the information differently.

うん、うん、the psychological effect this has is rather different from having the sentence "there if you need it". This is something I am of the opinion that some may have overlooked. While the actual format isn't very different, the way in which you process the fact/card is a lot different. From playing around with it last night, I can see this approach being especially effective in terms of using the target language to re-enforce the target language, in the more intermediate stage.



...which is of course to say, this is assumption based on my opinion.

Zarxrax Wrote:I think I've been doing something similar for ages now.
Only on my deck I don't have that cumbersome "clicking" step.

I simply have my vocabulary word displayed on the card in a large font, then the context sentence is printed right under it. When I review, I just look at the vocabulary word and try to figure out what it is, and then I am free to read the sentence for more context if I want.

I guess the only difference here, is in my method, the context sentence is already shown, wheras in your method, you have to click in order to get it to appear.
My method is probably a lot easier to set up and use, I think.

And honestly, I have never had any problem with "cheating" and looking at the sentence before I actually intend to. Maybe its just that my japanese level is still low, but I have to put quite a bit of effort into looking at a sentence before I can read it, unlike in english where I could read a sentence with just a quick glance.

My decks are also like this.
I have the word in a big-clear font and the sentence beside in a smaller font.
I read the word and if I know what it means I hit show the answer which triggers the audio for the sentence. Sometimes I guess the word wrong and realize only after listening to the audio, sometimes I need to read the sentence before remembering the meaning of the word and sometimes I can't remember even after reading the sentence.
This kinda split the hard work for judging how well I was in that card.
Easy -> Remembered without reading sentence (but the sentence will be read to me, so the context will still be sticked to my mind)
Good and Hard -> Remembered after reading the sentence. Good or hard will depend on the time I took to remember.
Fail -> Couldn't remember at all.

Anyway, I think I'm gonna give it a try for this 2-step method when I make a new deck for JLPT grammar... This might solve my bad habit of remembering the entire sentence instead of remembering what the grammar should be by itself.

Edit: I forgot to mention that all my cards have already sound from Core 6k10k and some from KO, and for new sentences I just use GoogleTTS or record them myself.

I just have a selection of several sentences on the answer side of my cards (example). It's unbelievably easy to make such cards en masse using cb's Epwing2Anki program.

Edit: pic