Back

[Kanji Koohii] Indentifying kanji instead of matching them to keywords

#26
(2017-03-02, 2:25 am)SomeCallMeChris Wrote:
(2017-03-02, 12:32 am)creamyhorror Wrote: Just to chime in on the issue, I think going kanji --> English meaning is fine. I'm not an RtK user, but learning purely to identify kanji is good enough IMO since you don't actually need to produce individual kanji, but words and phrases instead.

I understand where you're coming from, but honestly, you might as well not do RTK at all if you're going to go kanji->keyword. There are tons of kanji recognition methods out there.

The entire point of RTK is to be able to produce the kanji from the keyword, which is for two reasons... obviously, to be able to write the kanji - which actually matters quite a bit if you intend to actually interact with real Japanese people in the language, and is also supremely convenient for handwriting entry interface methods in cases where you aren't sure of the pronunciation of the word.

For most students though, the most important point is that the method makes the kanji extremely memorable as individual characters, in much the same way that writing words in any language is an extremely valuable tool for learning to recognize them. Of course you -can- also learn to recognize words (or kanji) without learning to write them, but learning to write them helps them sink into the memory very deeply, very quickly.

I absolutely agree that writing kanji is useful to cement them in your memory. I think it's a good exercise to do for a few hundred kanji (say 300-500), and would recommend it to learners. Once a learner's beyond that point, the choice of going recognition-only becomes viable, depending on what their goals are.

I should also mention that I'm agnostic to any method of learning Japanese, and look mostly at efficiency and effectiveness with respect to the learner's individual goals. Taking elements that others have recommended and experimenting with them is fine in my book.

Quote:If you're going to go the recognition direction alone, you might as well skip RTK altogether and just get to it with the 10x extra flashcard repetitions you're going to need to learn to read without writing.

I think if a learner practices writing basic kanji as I mentioned above, they'd be fine going on to learn to recognise further kanji without specifically writing them based on a keyword. It's not what RtK advocates, but it's a pretty efficient approach to developing reading skills. Certainly not on the order of 10x extra flashcard repetitions - it might even be faster since the learner doesn't have to practice recalling kanji from keywords for another 1,600+ kanji.

Once you get to a level where you can read texts and acquire new characters and words that way, I'd say it's pretty efficient to simply absorb characters as a side-effect of acquiring new words. I look up a word with a new character, look at the character information to see what other words it's used in, air-write the character if I think it's common, and then move on. I think many advanced people will attest to this, though of course we don't have scientific studies on which method produces the best results.

In summary I think
  1. Writing kanji is good practice
  2. Producing kanji from keywords isn't the only efficient way to progress
  3. Words and phrases and collocations are the levels that a learner should spend relatively more time at as opposed to characters or primitives
I also found an old thread here where the same debate occurred: http://forum.koohii.com/thread-11652.html anotherjohn made some good points in his reply.
Edited: 2017-03-02, 8:24 am
Reply
#27
Many (almost all?) of the Joyo kanjis are also meaningful words in their own right. Therefore when you memorize them you are memorizing vocabulary, and therefore, in my opinion, both kanji->keyword and keyword->kanji options should have the same standing in the system. The way I use Kanji Koohii is this: for every new word I look its meanings up in Jisho.org, and, if necessary, replace the Heisig keyword with this list of meanings.
Reply
#28
Hmm. That would be a ton of reviews though. I don't think it makes sense to have TWO SRS'es for the RTK study.

However it would make sense to add a vocab SRS. It was always on my mind but that's not something I could do now.

With a vocab SRS then yes, with many single kanji words you'd effectively review kanji > keyword. Until then I suppose a "free review" mode is better than nothing.

Problem is one of expectation and goal. The website 's main purpose is to help get through RTK. As such the method is about linking a absic meaning or placeholder word with the kanji writing.

When you replace the keywords I assume you are just looking for a better one? But keep in mind you'll need a unique "hook" for 2000+ characters. or did you mean you use a Japanese word as the keyword? (which is something quite a few people like to do)
Edited: 2017-03-02, 9:35 am
Reply
August Sale (14th - 25th): 30% OFF Premium PLUS - 25% OFF Premium
JapanesePod101
#29
If by generalizing the keyword field to include several meanings two or more kanjis turn out to have the same meanings, then so be it. In practice, I doubt that this is a very common phenomenon, and when it does happen the group of synonymous kanjis is probably small (two?), so for all intents and purposes you are still left with a unique (generalized) keyword for every kanji.
Edited: 2017-03-02, 9:52 am
Reply
#30
There is a new "Kanji to Keyword" checkbox on the Kanji Review page on the staging site ( staging DOT koohii DOT com ).

Please continue feedback if possible in the follow places:

Github issue
Add a Kanji > Keyword review mode #79

Gitter (chat)
https://gitter.im/kanji-koohii/feedback

PS: it actually works with example words. However it's not the idea I suggested above, it' basically just the kanji review reversed, so sometimes you get 2 or 3 kanji compound words. So you can see example words if there are any, and the meanings are shown after flipping the card. Some day, I'd like to let users "favorite" example words, and provide good choices by default, instead of the random ones. However for now I hope this is usable.
Edited: 2017-03-02, 12:05 pm
Reply
#31
Fabrice, I just wanted to say thank you for all that you do. I am one of those that does my actual SRS reviews with the help of an anki deck (usually manually copy-and-pasting stories from koohii when I need them) - I have tried to do SRS reviews on the site but found that anki just works better for me - but the mechanism on the site for allowing people to share, upvote and modify each other's stories is absolutely invaluable and I don't know what I would do without it. I do try to participate in the upvoting and creation of stories as often as I can. And the forum site is, if anything, even more valuable; there is such a wealth of shared knowledge about Japanese language study here. I hope that you know that even those of us who don't directly do reviews through the site appreciate it more than you can ever imagine.
Reply
#32
I've just given the new feature a ride. Looks great. Thank you!
Edited: 2017-03-02, 12:23 pm
Reply
#33
Just to avoid confusion. Did you mean you wanted ALL your SRS reviews to be kanji > keyword?

As far as I'm concerned I don't care. I don't know if I want to add an option for it in Account Settings but the option is there, undocumented. If you add "&reverse=1" to the SRS URL for example for due cards: /review?type=expired&reverse=1
Reply
#34
Quote:Just to avoid confusion. Did you mean you wanted ALL your SRS reviews to be kanji > keyword?



I meant that I would like to be able to choose whether I want the review to go keyword->kanji or kanji->keyword. The two options should be thought of as separate collections of flashcards with separate SRS counters.
Edited: 2017-03-02, 3:52 pm
Reply
#35
Hi, Fabrice. Sorry to be somewhat hijacking the thread, but after reading some of the posts I just felt the urge to say a big "thank you" once again for all your efforts through this years. I'm the opposite case to tanaquil: discovered your site when I was like 500 kanji deep into Heisig's method and started using the shared stories; a little later ditched the crappy software involved with my reviews and tried Anki, but never really got the fever. Since then I've been gladly using your site's SRS for years! So it is extremely valuable to me.
I finished RTK3 almost 3 years ago but never found the time or the motivation to jump to a vocabulary deck seriously. Tried Anki again with an optimized core 10k, still too dry an application to my tastes and recognition cards weren't working for me. So the only thing I've been doing Japanese-related lately is watch some subbed anime from time to time, come around this ever kind and supporting forum, and use your site's SRS on an almost-daily basis (which takes basically just a few minutes, even if I skip a day or two, since my daily review count is almost always below 20).

I obviously wouldn't have made it that far, with so little effort, in this mostly pointless hobby of mine, if it weren't for ALL of your site's features. I would have abandoned it a long time ago...

(...I just secretly dream sometimes that you'd also do a vocabulary sister site at some point: with a production centric SRS and where users could share their explanations on nuances and usages of words and vote for them... but I know that'd be a daunting, discouraging task for just one man -and I don't even have a definite picture of what it would look like- doing it all alone for no money and lacking any clear personal objective at the end of the road, so please don't take this as a request of any kind ;-). I guess I'll be using, in the near-to-middle-term future, either memrise or the delvinlanguage thing someone posted recently -which I tried and found it to be really great, but it has a crappy SRS scheduler and an still incomplete database-... In any case, I'd be so much eager to try something that came directly from you... I guess I'm just a cry baby ;-))




@OP: As some others have already expressed here, I have serious doubts about the efficacy of what you're trying to do... but as you clearly stated, you're a grown up and seem to have thought properly about it, and I must confess I'm curious. So if you really push forward, it'd be great that you made a "study log" thread to follow ;-). Good luck!
Reply
#36
Thanks for the kind words faneca!

faneca Wrote:(...I just secretly dream sometimes that you'd also do a vocabulary sister site at some point: with a production centric SRS and where users could share their explanations on nuances and usages of words and vote for them... but I know that'd be a daunting, discouraging task for just one man -and I don't even have a definite picture of what it would look like- (...)

Vocabulary is something that was always on my mind. Let's just hope I can get some help once I open source (edit: I know how to do it, and how to design it, with contributions I could focus my efforts on such new features). I just need . Last time I tried to open up the code oddly enough even though the repo was on Github I didn't use Github's Issues. Now maybe it will work better since I can delineate clear tasks, from very small to medium complexity, that people can chose. That plus being more available via Gitter etc.
Edited: 2017-03-03, 9:18 am
Reply
#37
(2017-03-02, 12:00 pm)ファブリス Wrote: There is a new "Kanji to Keyword" checkbox on the Kanji Review page 

Please continue feedback if possible in the follow places:

Github issue
Add a Kanji > Keyword review mode #79

Gitter (chat)
https://gitter.im/kanji-koohii/feedback

This is live now, and the option was added to the second Kanji Review mode. Quoting news post:

Quote:The most convenient option likely is with "Review from learned kanji". (second mode in Kanji Review page). After you review with the SRS, you can revisit those same cards (due and scheduled) in "kanji to keyword" direction in the free review mode (thus not affecting the prior SRS review scheduling).
Reply
#38
I'm not sure how helpful it would be to do kanji --> keyword when it comes to writing kanji. But there's also a huge difference between writing kanji and writing words.

If we're talking about basic kanji where they're basically the word itself, it may not be that hard. Words such as 車, 君, 私, 俺, and 心. These are extremely easy to recall.

However, once you get into kanji compounds, RTK alone just doesn't cut it. You may be able to recall certain kanji just for what they are, but I think kanji --> "possible meanings" may be more efficient when it comes to making out the difference between similar kanji, which might actually make it easier to recall kanji readings as well, since there's a lower chance you'll accidently mistake one kanji for another.

For example, today I was supposed to review 景, but I recalled 憬 instead. 景 usually means "view", whereas 憬 means "desire / to long for sth". But as we all know, it's normal to forget keywords after a while. What's more, I had けしき->?色  written on the front to help me recall the kanji, but it still wasn't enough to recall the specific one I needed even though I know both of them (景 is obviously part of 憬, so it's not like I can't know it).

Other kanji such as 疑 and 擬 are obviously different, but can I naturally recall their meanings if I see them in a new words in order to help me infer the meaning? In the case of 擬, I'd have to remember the story, but even then it probably wouldn't be very helpful. But if you're used to going from kanji to "possible meanings", I guess that wouldn't be much of a problem.

Perhaps a better example is 斎, 剤, and 済. Writing them is not exactly a problem if I can recall the exact one I need. But I don't remember the meaning of any of them. I thought 済 was the one used in detergent, that is to say 洗剤, but it's pretty clear that's not the case. I know the kanji, but not that meaning. If you go from kanji to meaning and pay close attention to the primitives, I supposed that wouldn't be as much of an issue. This skill would come in handy both in reading and writing.

Frankly, I don't know how high the cost would be. But as it stands right now, I can write a lot of kanji, but if you ask me to write words... it'll be a disaster. Probably not much different than if I'd studied things the other way around. So at the end of the day, if you end up having to go through tons of production flashcards or lots of writing to be able to recall the correct kanji for the words you want to recall no matter which path you take, is it really safe to say that keyword --> kanji is "the way" to go?

(But then again, Japanese is so complicated that no matter which way you go, you still have tons of readings to learn, words that you won't be to recall even if you know their meanings (e.g. 印象), etc.)


Anyway, I'm not sure if I should be posting this here, just trying to say that it's nice idea to have this option as going from kanji to meaning does seem to have its merits to me.
Edited: 2017-03-03, 3:48 pm
Reply
#39
I don't think there is a right or wrong way to review kanji.
Either seems valid. Either have their drawbacks.
I mostly have reviewed kanji going from kanji to keywords because it seemed the most natural to me.
By doing that way, I don't feel bound to those words, you can almost forget about them and start to feel the meaning of the kanji. After all, the goal is to stop thinking in your own language.

Also I don't agree with Chris about almost everything he said, ^^; though I have nothing against him (I swear) 
That's funny because to me RTK is all about recognition.
That's how I understood the book and that's why I think it was good for.
RTK just does what its cover says: remembering the kanji (not knowing them).

I was producing kanji from keyword and keyword from kanji by using handmade flashcards way back before I knew about RTK (there was also no Anki at that time or jisho.org), and it didn't prevent me from forgetting about them. A real pain in the ass.
Kanji > keyword or keyword > kanji is irrelevant to the learning process, it's about the same: association of an abstract shape with a concrete word and using a silly story to make it sink into your brain.
But in the end, you won't rely on those keywords anyway. You will forget about them and hopefully, directly think in Japanese. Just use what suits you the most or do both and move on.
There are more important things to learn.
Reading RTK from cover to cover doesn't mean that you know kanji (nor kung fu).

It's just a first step in the learning process, though a big one.
It's up to you to add/refine as many knowledge as you can to those kanji/placeholders in your brain that you now remember.
So there no such thing as 'RTK is meant for...'; it is meant for remembering kanji like any other methods, but it focuses on simplicity over accuracy.
And I think there's a good reason for that.
It is because the different readings and meanings of a kanji only come into play when there is a context (i.e. when reading japanese materials). So there is no point on focusing too much on those things separately in a book, especially when you can't even remember their shape and basic meaning in the first place.
Though I agree that writing kanji should be part of the first step of the learning process, not because it is important, but because it is convenient to do it at the same time.
You can learn to write kanji no matter the way you review them.
Recognizing a kanji doesn't mean you know how to write it and vice-versa (that why many japaneses forget about kanji, but they still know how to read them), just don't fix your screen when the kanji pops up ^^.
Reply