Back

Returning to kanji.koohii.com after some time

#1
Last year, in preparation for the JLPT N2, I decided to learn all the Jouyou kanji using RTK and this site. That worked fine. I did the initial learning in around 6 weeks of full-time study and then worked to get all the kanji into the 5+ boxes at a slower rate while I did other study that I needed for the test.

I haven't kept up with revision on the site in the period between mid-November last year and around the start of March this year. As a result, I've got a backlog of around 1,000 kanji to review. While I would like to get back to using this site as part of my daily revision routine, I've got various problems:

  • the sheer size of the backlog
  • I want to scale back and primarily only study JLPT N5-N2 (and maybe N1) kanji (I failed the N2 by two points, so I want to take it again): I'd like to be able to defer or ignore the non-relevant kanji
  • my growing dislike of a lot of the keywords that Heisig uses
  • not being sure whether continuing with koohii is the right thing to do (keyword issues, and no way to learn readings and vocab in an integrated way)
  • general forgetfulness (after moving to a more vocab/reading-based approach, it's harder to trigger keyword-to-writing memory)
Although it's not needed for the JLPT, I do want to continue to be able to write kanji correctly so I do think that koohii can help with that. However, I don't want to spend an excessive amount of time trying to get back up to speed on the relevant kanji or doing a lot of housekeeping tasks like manually removing a lot of non-JLPT kanji or modifying keywords so that they show the JLPT level.

It's also quite frustrating having to deal with poor or opaque keyword choices, eg:
  • various ways to say "I" in Japanese: keywords like "private", "myself", "me" would be better as "I (neutral/private)", "I (young male)", and "I (coarse male)" or whatever, and "I" itself should be "I (numeral)"
  • demolish (壊) vs break (毀). The first kanji is a lot more useful, so it should map to the more common keyword "break (demolish)"
  • conceal (隠) vs hide (匿). The first kanji is a lot more useful, so it should map to the more common keyword "to hide (conceal)"
  • many other groups of synonyms and near-synonyms among English keywords
  • many English keywords that have multiple senses, only some of which apply in Japanese (eg, determine, recess, present)
  • no unified approach for kanji that render verbs like きく、さがす、みる、かく with different nuances
  • other minor improvements ("lightning bug" -> "firefly", "decameron" -> "tenday", 
  • opaque humble and honorific-style keywords (speaketh, doth, "do", "re-", etc.)
Then there's lack of integration with other aspects of learning Japanese. Specifically, learning kanji and vocab readings seems to be quite divorced from koohii's core feature, namely helping to learn and reinforce the keyword-to-writing part.

So, anyway, I'm just throwing this out there. How have other people dealt with these kinds of problems? Have you managed to get back into using koohii after an extended absence? How have you dealt with the backlog? Have you managed to come up with a more integrated approach (including writing, readings and vocab) that still uses koohii? Or should I just not bother keeping up with koohii and transition to other tools like Anki?

Any advice or experiences you'd care to share would be much appreciated. Thanks.
Reply
#2
(2017-03-16, 1:16 pm)declan Wrote: … there's lack of integration with other aspects of learning Japanese.

Yes, that's a problem with limitation of the Heisig system. Consider the various keywords for 偉:
  • RTK: admirable
  • Kanji Odyssey 2001: great, big
  • JiShop: great
  • Kodansha KLC: GREAT, eminent
The keywords impart some concept of "big" or "impressive" but won't necessarily make the meaning of compounds such as 偉業, 偉容, & 偉挙 obvious when you see them for the first time. (Maybe these aren't the best examples, but it's just an example.)

I would recommend using Anki to study compounds using this deck (and optionally this larger deck). Don't worry about the readings—just focused on how the kanji combine to create new meanings as ja-dark recommends here (with research-oriented commentary here). It will help solidify the various real-world meanings that each kanji have, and will ultimately make them easier easier to recall when you practice the writings.
Edited: 2017-03-16, 2:23 pm
Reply
#3
I can't comment on other limitations of the Heisig system as I usually review with my own anki cards, but I would like to +1 the request for some mechanism to help those of us who have fallen behind catch up. I started out on koohii many years ago and got to about 250, but my various attempts to recover reviewing on the site have all failed, in no small part because there is no mechanism for doing catch-up.

In anki, I can create two filtered decks that allow me to (1) review overdue cards, let's say half of which will fail (2) review all currently due cards, which will include both any that are coming due in their proper cycle, and any that are coming due because they were failed recently. Without an option like this, you end up with a morass where any cards that you review and fail end up back in the same pile as the backlog, and the backlog never ends.

Maybe this isn't so easy to code, though, in which case I guess I will stick with my anki solutions. It would be a really helpful option though.
Reply
EPIC SALE: Get 30% OFF Premium & Premium PLUS! (Sept 18 - 29)
JapanesePod101
#4
1. When a keyword isn't an unambiguous reference to a Kanji, you have to add clarifications, to make it so. You can also just change the keyword, if it isn't a primitive. But that doesn't seem necessary. The keywords aren't important, they don't need to be perfect.

Another thing that isn't important: knowing every single one of the 2000 Kanji. If you don't like a card, and you don't feel like coming up with a new one, or adding clarifications, just get rid of it. (you can do that easily in Anki...I'm not familiar with this site's interface, but I bet you can do it here too).
2.
Quote:Then there's lack of integration with other aspects of learning Japanese. Specifically, learning kanji and vocab readings seems to be quite divorced from koohii's core feature, namely helping to learn and reinforce the keyword-to-writing part.
I disagree that the system is divorced from learning Kanji. That IS what you're learning: the Kanji. Once you are able to write the Kanji based on the keywords, you should also be able to recognize them. At least I haven't heard anyone say they don't, before.

It IS of course divorced from readings. But that's by design. It's not meant to teach Japanese, it's only meant to ease westerners into a SE Asian approach to writing that is completely alien to them.

Which brings me to how I transitioned from that to actual Japanese: I never tried to study readings. I followed the advice from AJATT, which was to just dive straight into sentences after RtK, without worrying about readings. So don't worry about trying to "integrate" RtK with Japanese. Just dive into Japanese, and let RtK work for your subconsciously. It DOES help your progress.

3. Concerning the size of the backlog: this is a good place to apply the 80/20 rule, by switching to RtK Light (which is a collection of about 1000 of the more essential Kanji), or the even smaller collection of 500 absolutely essential Kanji. (Note: I'm not suggesting that you'll get 80% of RTK's benefits from learning 500 Kanji...but you'll definitely get a lot more than just 25%...more than even 50%, imo.).

Again: I don't know this site's interface, but you can do everything I suggested in Anki. In the case of no. 3, you'd have to start with a fresh deck. But it's a lot better to start over than to try and catch up (even if you decide to start over with the full 2000 Kanji set again). In fact, very little of your previous work would go to waste, by starting over.
Edited: 2017-03-16, 4:13 pm
Reply
#5
@tanaquil Cards that were failed recently I assume would be in the lower piles with Koohii's SRS. Cards that you know better would be on the last piles. Some users prefer to filter due cards by pile. To do that you click directly in the SRS chart. Asuming you add new cards, this is how you'd whittle down due cards reviewing a long time ago, and due cards from recent reviews.

Other than giving more fine grained options I don't see how you can solve a backlog. The recommendaiton is always the same

1) dont reset the SRS, even if you pass 10% of the cards its better than starting from zero
2) decide on a daily limit and stop worrying about it. If you have a backlog, piroritize the due card piles at the lower end (recent additions) or higher end (likelier to pass due to multiple success reviews over weeks and months).

Heisig keywords is another thing entirely. Koohii 's primary goal is to help someone get through RTK.

If you want to significantly alter RTK's keywords or system, then it makes sense to use software like Anki with whichever custom deck suits your needs.
Reply
#6
(2017-03-16, 3:25 pm)tanaquil Wrote: I can't comment on other limitations of the Heisig system as I usually review with my own anki cards, but I would like to +1 the request for some mechanism to help those of us who have fallen behind catch up. I started out on koohii many years ago and got to about 250, but my various attempts to recover reviewing on the site have all failed, in no small part because there is no mechanism for doing catch-up.
I 100% guarantee you that the absolute best mechanism for doing catch-up is to start over. And I don't use this kind of language often. In fact this might be the first time I ever "100% absolutely guaranteed" anythingSmile.

But that's how certain I am that trying to catch up is not worth it, and starting over is always better, when it comes to SRS reviews that you've fallen behind on years ago.

And don't worry, the work you've done previously won't go to waste. The knowledge you've accumulated is still there, deep down in your memory, and it will make your reviews much easier this time around.
Edited: 2017-03-16, 4:18 pm
Reply
#7
OK, so, when you fail any kanji, and then go through the process of restudying it, does it reappear as a green card (and eventually an orange card) in the #2 stack? I think that may not have been entirely clear to me. I thought maybe it only dropped down one level, say from 6 to 5.

(To be honest, the whole process of clicking on a failed kanji to restudy it and then eventually return it to the cycle is opaque to me. I usually muddle through it, but I can't say I entirely understand what the site is asking me to do. Half the time it seems like I click on the wrong button and have to start over.)

Stansfield: interesting, is it even possible to "start over" on koohii? To be clear, I am currently running an anki RTK deck that is up to 750 or so. You would think this would mean I would get a 100% correct rate on any kanji I retest on koohii, but oh no, I always seem to fail at least one (probably one that was advanced out too far by my anki deck and would have been failed the next time I saw it). WTF brain. I could see how starting over might actually be less demotivating for me.

Maybe it's a moot point - I am pretty committed to my anki deck, and to be honest, I'm not sure I have the time to maintain regular reviews on koohii as well - but I want to support koohii and the idea of having the full range of stories right in front of me when I want to review a forgotten kanji is attractive. Maybe one of these days I'll catch up.
Reply
#8
(2017-03-16, 4:06 pm)Stansfield123 Wrote: I disagree that the system is divorced from learning Kanji. That IS what you're learning: the Kanji. Once you are able to write the Kanji based on the keywords, you should also be able to recognize them. At least I haven't heard anyone say they don't, before.
FWIW, I couldn't. I went through RTK and could reliably do keyword-to-kanji, but that was not of any help in getting keyword from kanji. If I wanted that I think I'd need to srs it separately.
Reply
#9
(2017-03-16, 5:08 pm)pm215 Wrote: that was not of any help in getting keyword from kanji
That's gotta be hyperbole. Not of ANY help? Surely, it at least helped SOME? You didn't even recognize 一 and 二 , after going through RtK?
Edited: 2017-03-16, 6:33 pm
Reply
#10
@tanaquil

The Leitner SRS implementation used to be pretty harsh: a failed review ("No") sends the card back to the Study pile (red cards).

Some time ago I added "Hard". This is the one that sends cards down one pile, with lower intervals. Hence eventually they find themselves back in pile #2. Hard answers however NEVER go to restudy and will always stay in pile 2 until you pass them.

edit: So to be clear, it is possible to entirely skip the Restudy process if a user doesn't want that. This is probably why also it makes sense to review piles separately. If you know you are reviewing the last pile, you'll probably want to say "Hard" instead of "No".

Adding cards to "learned list" from the Study pages and then succesfully reviewing them does send them to pile #2, yes. (ie. one review). So when they are first due, they are orange cards in pile 2.

No point in using two SRS. Use whatever works for you.
Edited: 2017-03-16, 7:03 pm
Reply
#11
(2017-03-16, 6:29 pm)Stansfield123 Wrote: That's gotta be hyperbole. Not of ANY help? Surely, it at least helped SOME? You didn't even recognize 一 and 二 , after going through RtK?
I recognized those already since I already knew the numbers, so RTK didn't give me anything new there. But looking at a character typically my brain can dredge up either a Japanese word it's used in, or a guess at the on-reading, or a vague area of meaning, because for all those I've either anki'd that or have absorbed it through sheer exposure. (Or else just a vague "I know I've seen this before" feeling of familiarity.) What I can't do is tell you the RTK keyword, and that seems unsurprising to me because I've never tried to memorise the association in that direction.
Reply
#12
(2017-03-17, 3:49 am)pm215 Wrote: I recognized those already since I already knew the numbers, so RTK didn't give me anything new there.

What about 愛、女、子、人、etc. RtK didn't help you recognize those either? You can write them all, based on their keywords, but when you see them, you don't know the keyword?
Edited: 2017-03-17, 5:45 am
Reply
#13
(2017-03-17, 5:43 am)Stansfield123 Wrote:
(2017-03-17, 3:49 am)pm215 Wrote: I recognized those already since I already knew the numbers, so RTK didn't give me anything new there.

What about 愛、女、子、人、etc. RtK didn't help you recognize those either? You can write them all, based on their keywords, but when you see them, you don't know the keyword?

But those are incredibly common to begin with, so isn't it kind of hard to tell how any system helped with those? If anything I'd suspect RtK to help with the more complex kanji because they are composed of more parts. When I started going through the Core decks but didn't bother writing or really learning kanji individually at all I would recognize many words just by a rough shape (so I might recognise 噂 by "it's only one kanji" with the mouth on the left and "many strokes" on the right side or something...) I'd imagine learning with RtK helps prevent that somewhat... Doesn't really matter in the end if you really track back to the RtK keyword I guess?

edit: I feel bad for not saying anything on topic... Uh... I'm an on and off learner, so by now I'm pretty used to tackling huge piles of backlogs in any form of srs system I used... For me just doing a set amount of reviews each day has worked pretty well, but yes it will take some time until the backlog is cleared and that can be frustrating at times. (I think it took me roughly 3 weeks to get my iknow backlog down to zero this time around, but I was still a bit inconsistent the first few days and I already kept adding a few new cards during the end phase of this, just because~)
Edited: 2017-03-17, 6:18 am
Reply
#14
(2017-03-16, 6:59 pm)ファブリス Wrote: @tanaquil

The Leitner SRS implementation used to be pretty harsh: a failed review ("No") sends the card back to the Study pile (red cards).

Some time ago I added "Hard". This is the one that sends cards down one pile, with lower intervals. Hence eventually they find themselves back in pile #2. Hard answers however NEVER go to restudy and will always stay in pile 2 until you pass them.

edit: So to be clear, it is possible to entirely skip the Restudy process if a user doesn't want that. This is probably why also it makes sense to review piles separately. If you know you are reviewing the last pile, you'll probably want to say "Hard" instead of "No".

Adding cards to "learned list" from the Study pages and then succesfully reviewing them does send them to pile #2, yes. (ie. one review). So when they are first due, they are orange cards in pile 2.

No point in using two SRS. Use whatever works for you.

Thank you, I didn't realize Hard worked like that! I was puzzled when the system kept listing "hard" kanjis as "failed" in the review list at the end. I tend to overuse Hard, it's a weakness of mine. I overstudy a lot.

I'll have to make a point of using hard less and reviewing the stacks at the bottom more. That "due" number when I drop in to look at various stories (which I do almost every day) bugs me, it's an OCD thing.  Big Grin

Thanks for all that you do for the site.
Reply
#15
@tanaquil

I see it can be confusing that Hard answers are counted as failed reviews.

The SRS review summary should probably have labels indicating the actual answer used, in addition to the raw "pass" "fail" counts. I'll add an issue for that.
Reply
#16
(2017-03-17, 6:04 am)sumsum Wrote: For me just doing a set amount of reviews each day has worked pretty well, but yes it will take some time until the backlog is cleared and that can be frustrating at times.
It's simply this. You don't need a complicated system and you don't need to finish the entire backlog in one day. No matter how many reviews you do in a day whether it's 20 or 200 is getting you that much closer to catching up. Keeping at it and eventually you will catch up if you aren't adding any new cards.
Reply
#17
I had a huge RTK backlog (in Anki). Decided to revive the deck spending the least effort possible, by keeping only remembered and almost-remembered cards. I set the leech threshold to one failure - you can probably do the same on Koohii just by not doing anything with the failed pile?

Then when reviewing:
  • If I got the card right I picked "good".
  • If I didn't get it, but the story came back instantly when I saw the kanji, and it also seemed like a useful kanji, I picked "hard".
  • Failed (i.e. suspended) everything else.
Starting over wastes time on the cards that would still be remembered. But after a long break, I suspect it's more efficient to bulk-relearn all the forgotten cards in RTK order, rather than randomly as they come up for review. I haven't actually gone through my dead pile and might never do it, but that's what I'm thinking if I do.
Reply
#18
sumsum Wrote:But those are incredibly common to begin with, so isn't it kind of hard to tell how any system helped with those? If anything I'd suspect RtK to help with the more complex kanji because they are composed of more parts. When I started going through the Core decks but didn't bother writing or really learning kanji individually at all I would recognize many words just by a rough shape (so I might recognise 噂 by "it's only one kanji" with the mouth on the left and "many strokes" on the right side or something...) I'd imagine learning with RtK helps prevent that somewhat... Doesn't really matter in the end if you really track back to the RtK keyword I guess?

It's not as simple as "tracking back to the keywords doesn't matter". In some cases it matters, in some cases it doesn't. In some cases knowing the keyword helps a lot, in some cases it helps a little, in some cases it doesn't help much at all.

With all the Kanji I listed, knowing the keywords helps a lot (because they're common, concrete keywords that are accurate translations of Japanese words/morphemes written with those Kanji). And there are tons of such keywords in RtK. With more abstract keywords, it helps a lot less, and sometimes none at all (because the keywords are less accurate).
Quote:But those are incredibly common to begin with, so isn't it kind of hard to tell how any system helped with those? If anything I'd suspect RtK to help with the more complex kanji because they are composed of more parts.
RtK is supposed to be done up front, before you study any Japanese, to familiarize beginners with an alien writing system. It's a lot less helpful for intermediate or advanced students, who already know some Kanji. As such, its main benefit is in teaching the most common Kanji, with the most concrete keywords (irrespective of the Kanji's complexity). That's what beginners need.

And most of those Kanji are fairly simple, But not all, some are complex. 愛、for instance, which I listed above, is a complex Kanji. But it has a simple keyword, so I find the claim that someone could write it, but not know the keyword for it, bizarre. That's why I listed it.

As for the Kanji with the abstract, ambiguous or confusing keywords...RtK doesn't really help much with those. I guess you can learn to write the Kanji, and figure out how they're used later, but that's a very fringe benefit. Beginners should just delete those cards. If you don't, that minority of cards will take up the majority of your time, and, in the end, learning them won't help you much anyway (because the keyword doesn't really tell the story of how the Kanji is used).

So, if you apply the 80/20 rule twice, like this:
1. get rid of 1000 Kanji that are not that common (and aren't used as primitives for common Kanji), by going with the RtK Light deck (or the 500 Kanji deck...not sure what it's called, sorry).
2. on top of that, get rid of any Kanji with bad or confusing keywords (either up front, or once it becomes clear that a card is giving you trouble)

....you will have cut down on the majority of your workload, but will still gain most of the benefits of doing RtK. And, again: RtK is for beginners. If you already know a lot of Japanese, it's ridiculous to start studying Kanji with English keywords.
Reply
#19
(I feel kinda bad that we've hijacked this thread, so just one last post on this tangent.) I went through RTK as an intermediate student largely because I wanted to be able to actually write the kanji without a huge memorization and repetition burden. (IMHO if all you care about is being able to read them then almost any approach will work and RTK is way overkill, and in any case you'd be better off memorizing kanji->keyword or meaning, not keyword->kanji). Anyway, it worked in terms of being able to write them from keywords, but if I can't go from kanji to keyword then I can't do 'entire word -> kanji' and being able to write a kanji from its keyword is a useless party trick. (I have a currently-on-hold project to go through RTK again with Japanese prompts, which I got up to frame 600 or so on. That might or might not work better.)
Reply