Of course, many Chinese characters where originally built with a phonetic component, and sometimes (but not always) this carried all the way to modern Japanese.
Notice how many of the kanji containing 方 have the onyomi ホウ for instance. Or how kanji containing 古 are often pronounced コ. Or kanji containing 正 are often pronounced セイ or ショウ (or sometimes both). There are so many more. Just click on the radical pages on WK and see if you notice a reading coming up more often than the others. Take 義 for instance, when you see a radical like that you know that you’ve hit the motherlode.
It’s not 100% of course but it helps a lot I find, and the more kanji you know the more you are likely to notice these patterns.
It even helps guess the pronunciation of unknown kanji, for instance I managed to guess how to pronounce 故障 the first time I encountered it in the wild because I identified 古 and 章 and guessed that it had a good chance of being こ and しょう.
Of course that only works sometimes, but it can be helpful to look up unknown kanji without dealing with a radical search or OCR.
I talked about this a lot in my level 60 post, but I agree! I think the structure of WK is very effective, and I certainly wouldn’t have gotten anywhere without it, but for the last half of the program, I was relying far more on the semantic-phonetic composition script to memorize kanji readings, especially when WK would try to work like three or four radicals into a mnemonic when it could have just used the actual kanji component instead. I wish they’d consider incorporating the keisei script information into the vanilla program and tweaking the order that things are taught in order to reflect that.
Yeah, I love the whole concept of semantic-phonetic composition, and at this point it’s actually helped me countless times in recognising unknown kanji or remembering. The semantic component is also pretty interesting, although often harder to work with - but there have been a decent amount of times where I’ve been able to guess both the meaning and reading of certain kanji from seeing these alone, which is awesome. The semantic ones tend to be harder to spot compared to their original though.
Yeah I have less luck with the semantic side but there definitely are a few worth noting like 疒 for health conditions and 月 (as a modified form of 肉) for organs/body parts. Those are fairly reliable, although for 月 there are many other uses of course. 火 also tends to have rather literal uses in my experience (unlike 水/氵which seems to be used for every other kanji to mean a billion different things).
And let’s not talk about bloody 糸. Ancient Chinese scribe: “hey you want a new character? How about I put 糸 in there real quick? Looks baller bro.”
Oh yeah, the fact that WK doesn’t even tell you that 月 means “body part” in most Kanji is just … weird? Why come up with stupid moon mnemonics instead?
It definitely feels like a lot of radicals exist just to gate content. In fact, according to this post something like 323 radicals (or about 2/3 of all radicals) had corresponding kanji on WaniKani. That’s about two full levels worth of redundant items.
While I do like the level-system, I don’t think it justifies that kind of busywork.
I wish they introduced the new radicals directly with the first kanji using it instead of separate SRS entries, and then force a minimum one week timeout on level up (regardless of kanji completion) to prevent accidental fast levels. This way you could pace your lessons any way you like without having to worry about rushing the radicals (if you try to do one level/week like I do) but WK would still control the max pace.
I think it would make WK’s content managers’ life easier too since they would have a lot more flexibility for moving things around and introduce/remove radicals as they see fit.
I definitely agree duplicating kanji as radicals and on top of this sometimes giving them different main names is not a good idea. This feels like leaking implementation details to the UI. If the kanji composition system requires some kanji to feed into other kanji, that’s fine. However, these extra radical/kanji entries shouldn’t go into SRS at all.
One way I would like to solve this in my app is by not having radicals as a separate item type at all. Some kanji will be quizzed for reading, some not.
Also, phonetic-semantic composition is noticeable enough that people figure it out eventually. It’s worth utilizing it, therefore!
Maybe introducing every new combination of kanji as a radical would lead to too many radicals, and that’s why they don’t introduce more “combination” radicals.
Having all kanji be available as mnemonic components for other kanji is the only thing I liked in RTK over WK, so some of the people who want to learn that way may be interested in using a more RTK-based site like kanji koohii.
