I've been wondering this lately. I finished RTK at the very end of last year and many of the keywords don't help me that much anymore. At one point, I began to devise several clues to help me in my reviews, because keywords can be very vague sometimes and I didn't have actual Japanese words to use as a hook.

I'm not sure how to deal with certain failures at this point. For instance, I've just "failed" the card 路 (path) because I remember 露 (dew) instead. 露 is a more complex kanji and even has 路 as a primitive, so I clearly haven't forgotten it - at least not as a primitive. But because I don't have Japanese keywords for all kanji yet, it takes much longer to review kanji than vocabulary in general. Maybe I should just press whatever (hard, good, easy, I don't know... would you recommend doing this? and if so, which one?) and hopefully I'll have learned a Japanese keyword by the time it pops up again in my reviews?

I thought going out of my way to learn words with these kanji might be helpful, but then again I don't know if they would stick since I'll most likely end up not even seeing these words that often.

It seems pretty clear to me that 路 and 露 are different kanji with completely different keywords, so if you are confusing them, i'd say it's a solid fail.  When I might pass something is more like 療 (heal) and 癒 (healing) because the meaning is so close, they might as well be interchangeable.  I figure once I learn words that contain those two kanji, the differences will sort out themselves.

What I've been doing is marking all of the kanji that I get confused and then studying them in another deck.  This thread is a good starting point for making a list and this thread is a way to set up your deck so that the algorithm keeps lookalikes together.  You can suspend the cards that don't give you trouble and set the others to relatively long intervals.

I (fairly) strictly fail cards that are not correct. And also (try to) use the three pass buttons consistently.

Like you, my vocab decks are less painful than my RTK decks. I suspect partly because the RTK keywords are a bit vague and partly because the vocab deck does not have as many obscure kanji.

Some suggest changing from English to Japanese keywords might help.

Another alternative is to look at the kanji you are failing the most and figure out why. This thread is helpful at least to see where others are having issues:

http://forum.koohii.com/post-205380.html#pid205380

I don't know. Sure, the keywords are not the same... but does that even matter? While remembering specific keywords may be oftentimes handy when you're learning new words, they're not actually Japanese words. For instance, it's easy to remember 私 even if you don't remember that it also means "private" in certain compounds.

I feel cheated whenever this happens. If I know that 雨 and 路 are used as primitives to make up 露 and other kanji, then obviously I know they can also be used separately as individual kanji as well. I don't know, maybe one will think the correct kanji to a keyword is 察 by accident, but that doesn't mean this person has forgotten that 祭 can be used on its own. It just means this person has recalled more information than necessary, which I supposed shouldn't be an issue unless you do that when I write a Japanese words as well. But at that point, you'll have to associate a specific kanji with a specific word, which is when it really counts.

I feel this may be an inherent problem with keywords, because you're not trying to recall how to write specific words in Japanese. It's frustrating to have to fail a card that's supposed to be review again only next year, just because a keyword and clues are little too vague and I didn't know a Japanese word with that particular kanji last time I reviewed it.

(2016-04-28, 2:47 pm)FlameseeK Wrote: I feel this may be an inherent problem with keywords, because you're not trying to recall how to write specific words in Japanese.

Yes, I totally agree, but I feel if you don't retain the keywords without replacing them with something else, then you run the risk of forgetting more than just the keyword.

(2016-04-28, 2:47 pm)FlameseeK Wrote: It's frustrating to have to fail a card that's supposed to be review again only next year...

If this is the crux of your problem, you could just change your "new interval" in anki's lapses preferences.  I set mine to 15% so those failed cards with a 365 day interval end up with a 54 day interval instead of one day. Here is the Lapses section of the manual - the third paragraph talks about "New interval".

(2016-04-28, 2:47 pm)FlameseeK Wrote: I feel this may be an inherent problem with keywords, because you're not trying to recall how to write specific words in Japanese. It's frustrating to have to fail a card that's supposed to be review again only next year, just because a keyword and clues are little too vague and I didn't know a Japanese word with that particular kanji last time I reviewed it.

I still would fail anything that you fail, but also make a point of putting better hints on the front side or finding a Japanese keyword so that it doesn't happen again. It is frustrating to fail because of keywords that are synonyms, but when you write a bigger kanji with totally unrelated meanings in both keyword and Japanese words, that seems like a clear failure. Yes you clearly remember how to write it, but not when to write it.

I don't understand it when you say "remember how to write it, but not when to write it". I mean, it's not the goal of RTK to teach you when to write certain kanji. That's a part of vocabulary learning instead. However, the truth is I only do recognition vocab cards, so it may still be a good idea to use my RTK deck as an opportunity to work on production.

I don't know, maybe you guys rely a lot on mnemonics and keywords than me. When I was about halfway through RTK, I decided to use hints for pretty much every single new kanji, since keywords are sometimes confusing and unreliable (though some clues aren't that reliable either). It wasn't necessary to read them every single time, but the older the card, the harder it is for me to recall the kanji based on the keyword alone. I tried not to read the clues whenever possible, but I was always glad to recall the kanji even if a keyword didn't ring any bells.

I've never cared much about not recalling a keyword or story, because as many people say, they're just a means to an end - a quick way to input kanji into your brain - and that you're likely to forget them eventually. Except I feel that " eventually" has probably come quicker than in many other cases.

I've already started adding Japanese keywords (which is also boring because it slows down your reviews) to my RTK deck, but old cards are pain because they have neither have clues nor Japanese keywords. My learning kanji cards are as high as 97.77%, so I feel reviewing a relatively easy kanji from 0 because the hooks weren't good enough is a little wasteful. At the moment, I've only answered 78.3% of my mature cards correctly, which is less than ideal imo. So it really seems like a problem with the hooks, since I can easily recall them afterwards.

That idea of setting "new interval" to 15% sounds awesome. I've changed it and I feel it'll make kanji reviews a lot less painful, since I clearly don't need to review most of these old cards THAT often anymore. Maybe all I really need is that to make the switch to Japanese keywords less of a pain.

Thanks a lot for the input guys!

Adding Japanese keywords is a tedious process but I think you will benefit. I sure did when I added them Though I didn't remove the English keywords either because I still consider them as my primary source to try to recall the kanji and just mainly use the Japanese keywords as a way to not confuse them with others.

TD;LR: I use Japanese keywords further differentiate similar looking kanji.

(2016-04-28, 11:50 pm)RawrPk Wrote: Adding Japanese keywords is a tedious process but I think you will benefit. I sure did when I added them Though I didn't remove the English keywords either because I still consider them as my primary source to try to recall the kanji and just mainly use the Japanese keywords as a way to not confuse them with others.

TD;LR: I use Japanese keywords further differentiate similar looking kanji.

That's not a bad idea.  I think I'll do that.  Thanks!

Btw, it doesn't have to be tedious if you have a list of kanji and their associated keywords.  If the kanji is the first field, anki can update the existing notes with the new information.

Here is my current RTK format as I showed off in the anki color scheme thread. As you can see, both English and Japanese keywords are side by side. Kanji is isolated as the answer in a very obvious light orange.

(2016-04-29, 12:17 pm)yogert909 Wrote: That's not a bad idea.  I think I'll do that.  Thanks!

Btw, it doesn't have to be tedious if you have a list of kanji and their associated keywords.  If the kanji is the first field, anki can update the existing notes with the new information.

That's exactly what I did, add another field. It's actually easier this way. Most of the Japanese keywords I have in my deck is from jalup mod RTK but it doesn't cover all of the RTK kanji. I add them using a spreadsheet and importing. Though I am not entirely finished with adding Japanese keywords, I have enough for now in my deck and only add more when I hit kanji without them.

(2016-04-29, 12:17 pm)yogert909 Wrote: Btw, it doesn't have to be tedious if you have a list of kanji and their associated keywords.  If the kanji is the first field, anki can update the existing notes with the new information.

Yogert909,

Quick question - I would like to import Japanese keywords into my deck. I see that can be done by exporting the deck to excel, editing the file then re-importing back to Anki. This blog summarizes the process in a few dozen steps:

https://duolinguist.wordpress.com/2014/0...into-anki/

Is there an easier way you recommend?

(2016-04-29, 7:05 pm)scooter1 Wrote: Quick question - I would like to import Japanese keywords into my deck. I see that can be done by exporting the deck to excel, editing the file then re-importing back to Anki. This blog summarizes the process in a few dozen steps:

https://duolinguist.wordpress.com/2014/0...into-anki/

Is there an easier way you recommend?

I quickly browsed what this guy wrote. It seems like he's just writing down every single step. So it's really not as convoluted as it seems. You are really just moving tsv files back and forth between anki and excel. I don't recall ever having to open in notepad to change the text encoding to utf-8, but I'm usually using google sheets on a mac, so maybe things are different in excel on pc.

The key thing to remember is that anki has a function that it checks the first field to see it matches another card in the deck. If it finds a match, anki will update just the new information. You just need to make sure that mode is selected when you import and of course make a new field to receive the new information. And there's no reason you need to import information that hasn't changed, so if you are importing japanese keywords, I'd strip out all other columns besides kanji and japanese keyword in excel.

Let me know if anything is unclear.

There are 2 way you can use your exported .txt file: a) google spreadsheet or b) Open Office

Google Spreadsheet:

1. Have open a new untitled spreadsheet in Google drive
   
2. (In your exported .txt file of RTK) Select all and copy the entire contents of the file
   
3. Paste (CTRL V) into the spreadsheet. It will take some time (1-2 minutes) and you have yourself a spreadsheet!
   

OpenOffice:

1. Right click the .txt file and select "Open with.."
   
2. Choose "OpenOffice.org Calc" which is the equivalent of Excel for Open Office
   

Text Import Menu should popup

• Character Set: Unicode (UTF 8)
  
• Language: Japanese
  
• Separator options> Separated by> Checkmark "Tab"
  

You can see the preview near the bottom. If everything is in order, click "OK".

With both, just create a new column for where you want to place your Japanese keywords. Then save file as a .tsv file and you're ready to import!

Thanks Yogert and RawrPK - those steps are much simplified and very clear!

(2016-04-28, 1:12 pm)FlameseeK Wrote: When not to fail a flashcard?

When you know the exact answer, no doubt about it. Any other time, fail it. Don't even think about it, just hit one and move right along. Anki is all about failing cards. That's how you learn...by failing everything you don't already know.

(2016-05-05, 6:09 pm)Stansfield123 Wrote:

(2016-04-28, 1:12 pm)FlameseeK Wrote: When not to fail a flashcard?

When you know the exact answer, no doubt about it. Any other time, fail it. Don't even think about it, just hit one and move right along. Anki is all about failing cards. That's how you learn...by failing everything you don't already know.

I generally agree, but I don't think you need to be so perfect about it.  For instance, I mentioned I get 療 (heal) and 癒 (healing) confused.  I don't think I'm doing myself any favors by failing those two every time they come around just because I forget which one has the -ing.  There's a few other similar keywords that I let pass because I feel I know them well enough and the benefit I get from differentiating them further is not worth the effort it would take.  I will just learn the difference through actual use instead.

In truth, there are few 'exact, no doubt about' it answers for foreign language flashcards because there aren't many 1 to 1 relationships between most words between languages.  So I feel getting too fine grained in your definitions is a loosing battle.  I would rather spend my time getting a general idea of a word and then read that word in a bunch of different contexts to flesh out the meaning of the word rather than spending too much time re-learning to differentiate between (for example) sea and ocean, or lake and pond.

My test when passing cards is that my answer should be useful in actual practice.  So if I confuse 療 (heal) and 癒 (healing) while reading, I doubt it would cause me much problem in understanding what I'm reading, so I would pass that card.  On the other hand, using OP's example, if I mistook 路 (path) and 露 (dew) while reading, it's conceivable that I would have problems understanding, so I would fail that card.

(2016-05-05, 7:17 pm)yogert909 Wrote: I generally agree, but I don't think you need to be so perfect about it.  For instance, I mentioned I get 療 (heal) and 癒 (healing) confused.  I don't think I'm doing myself any favors by failing those two every time they come around just because I forget which one has the -ing.  There's a few other similar keywords that I let pass because I feel I know them well enough and the benefit I get from differentiating them further is not worth the effort it would take.  I will just learn the difference through actual use instead.

The problem with that is that if you always remember 'heal' and never 'healing' (or vice versa) then you're not doing yourself any favors by falsely passing the one that you never remember. Just fail it and put a clue on the front side. The obvious clues are "no little" and "no heart", but whatever works for you is fine.

To be honest, knowing what i know, I would want to change 癒　to 'soothe', but i can't recall if there are keywords that would conflict with that. Changing keywords after the fact can be a little tricky, but meh.　Sometimes it's just not worth keeping a bad keyword. Of course, by the time you know what to change it may be easier to switch to Japanese keywords which are pretty much unambiguous. Of course, there -are- words that are allowed to be various kanji of similar meanings so you want to be careful to avoid or disambiguate those; but for the most part, there's one kanji for a given word; so if you can't come up with the correct kanji for that word, you've clearly gotten it wrong.

In the case of  療 (heal) and 癒 (healing), this is where Japanese keywords would really benefit. Here is an example for both :

Production

Front:Heal 治りょう
Back: 療

and


Front: Healing いやし系
Back: 癒


Or Recognition

Front: 療
治りょう
Back: Heal

and

Front: 癒
いやし系
Back: Healing

I have no idea how the Recognition format I set up will work out as my RTK deck is production. Hope this helps! If anyone has some tips to tweak the set up, please don't hesitate.

(2016-05-05, 11:08 pm)SomeCallMeChris Wrote: The problem with that is that if you always remember 'heal' and never 'healing' (or vice versa) then you're not doing yourself any favors by falsely passing the one that you never remember. Just fail it and put a clue on the front side. The obvious clues are "no little" and "no heart", but whatever works for you is fine.

Actually, for these kinds of closely related confusing kanji, I always include both keywords in the answer (eg heal -or - healing).  So I actually have to know the correct answer is one of the two (and I usually think about which is my highest conviction answer).  Doing this, the keywords eventually sort themselves out and I can distinguish them over time.  

I should note that I am doing recognition, and would be much more strict if I were doing production considering that I wouldn't want to write an incorrect kanji.  However my desire to read and understand asap is much more urgent than writing and speaking, so I do recognition on most of my decks.

----

@RawrPK, that's a good idea and I almost did that when you suggested it last week or so.  But for the time being, I want to solidify the english keywords and make tracks on learning a targeted set of useful vocabulary.  Eventually I will learn vocabulary words for all of the kanji anyway, so I'm trying not to be so precious with each and every kanji.

(2016-05-05, 7:17 pm)yogert909 Wrote:

(2016-05-05, 6:09 pm)Stansfield123 Wrote:

(2016-04-28, 1:12 pm)FlameseeK Wrote: When not to fail a flashcard?

When you know the exact answer, no doubt about it. Any other time, fail it. Don't even think about it, just hit one and move right along. Anki is all about failing cards. That's how you learn...by failing everything you don't already know.

I generally agree, but I don't think you need to be so perfect about it.  For instance, I mentioned I get 療 (heal) and 癒 (healing) confused.  I don't think I'm doing myself any favors by failing those two every time they come around just because I forget which one has the -ing. 

When you don't have confidence in Anki being able to help you learn a card (like you clearly don't, in this instance), then the correct option is neither passing nor failing the card...the correct option is suspending it.

But I don't think you can always be aware that Anki isn't working, like you are in the above example. Plus, it takes a lot of time and energy to think about whether Anki is working or not...time and energy that could be spent on learning Japanese.

That's why there's a fail safe built into Anki, that takes care of the problem without you having to worry about it: it's the leech threshold...but it only works if you stick with the plan, and fail cards whenever you don't know the answer (and set a really low leech threshold...definitely under five).

P.S. In your specific example, adding a clue would also solve the problem. But there are plenty of cases where Anki fails for other reasons, not just because the keywords are confusing. That's why the only reliable method to weed out these cards is the leech threshold...but you have to fail cards, you can't cheat, for that to do its job.

Suspending might be a good call possibly. Suspend one of them and when you're at a point where that particular card is mature, you can then unsuspend the similar card.

Though it might be a a longer process but IMO wouldn't it be also as long dealing with them both? I think it's an option to consider.

(2016-05-08, 4:31 pm)Stansfield123 Wrote: When you don't have confidence in Anki being able to help you learn a card (like you clearly don't, in this instance), then the correct option is neither passing nor failing the card...the correct option is suspending  it.

This is clearly an example of perfection being the enemy of good enough.

My intermediate goal is to be able to read. Anki has helped me recognize 2 kanji that have very similar keywords. True, currently I am not confident that I can determine which of the two keywords is the correct one. But I don't see it as much of a problem for thee reasons. 1) keywords often don't match every meaning of the kanji anyway (especially in compounds) - in those cases keywords are used as a memory hook to learn a vocabulary word. 2) in cases where the keyword is synonymous with the meaning of the kanji, the difference between heal and healing is never going to cause a problem in understanding while reading. 3) With actual use, I will naturally become familiar with these two kanji without any additional study - so why bother.

TL;DR I don't need to waste time being perfect when I already know these two kanji well enough to be useful for my goals.

(2016-05-08, 7:25 pm)RawrPk Wrote: Suspending might be a good call possibly. Suspend one of them and when you're at a point where that particular card is mature, you can then unsuspend the similar card.

Though it might be a a longer process but IMO wouldn't it be also as long dealing with them both? I think it's an option to consider.

I've already been down this route and it's not a bad one, but I'm not so sure how good it is either.  Even mature cards can become confusing once you add a cards that looks a lot like it.  This happened to me about 6 months ago when I un-suspended several hundred leeches.

I'm actually not sure that suspending leeches helped me in the long run.  Suspending leeches helps somewhat when you are only concerned with the raw number of facts learned.  However if you are concerned with learning a specific set of facts, it may be desirable to just power through it.  

The main problem with suspending leeches (and to a lesser extent, manually suspending) is that you are already so far into the learning process that suspending won't save time compared to continuing.  In other words, for facts that you absolutely want to know, suspending will be beneficial only if studying the card at a later date will take less time than continuing to study it now.