Inefficient radical combinations

When I was taught 時, 待 and 持 I was annoyed that the right half of each was treated as “grave + measurement”, making the mnemonics rather complicated, overly similar, and far-fetched.

I figured 寺 might mean something on its own, and I looked it up on Jisho, and lo and behold: it means “temple”! So I could just remember that, and then invent my own way easier mnemonics for 時待持. It’s nicely compositional and sensible that way.

A few levels later, I learn the wanikani radical 寺 (Temple). I was a little annoyed, because I wanted that taught to me sooner; so badly that I learned it myself!

Similarly 客 gave me trouble — the mnemonic is really bizarre because it has to combine too many things into one: “holding a winter helmet in your mouth, blah blah blah”, what? — and now I know 各 (Kiss), and I can simply remember 客 as “oh, you greet a guest with a kiss on the helmet”.

I just want to know, is this something other people have had trouble/frustration with? It feels like such an obvious imperfection in the whole “building blocks” system. It’s not a huge problem, but I would really prefer to know about these cases ahead of time. Can you folks think of other examples?


Cases I can find:

is ‘grave + measurement’ in 時 待 持
is ‘winter + mouth’ in 客
is ‘grave + pile’ in 室 屋
is ‘sword + mouth’ in 招
is ‘sword + drop’ in 認
is ‘rice paddy + treasure + ground’ in 便

(I looked at the radical list for stuff that looks suspiciously “compositional”, and then used the API to find cases where not that radical but its parts are used in a kanji breakdown. There are almost definitely cases I missed!)


It happens sometimes, a quite bad example is , which got five radicals instead of two with a kanji you already know (疑). I also dislike that a kanji is not considered to use its own radical, so it is not so easy to look it up.

I would be careful with “obvious”, if you juggle around 2000 kanji it’s not that obvious, maybe they changed the order and thought the kanji/radicals are not available (at that particular level) or something.

You can install my semantic-phonetic script, it has better coverage than WK if you like “kanji split only once analysis”. But from that I can tell that it’s not easy to find good splits.

Finally, I think they are working on the radicals, I expect a better coverage and lesser splits afterwards.

A bit unrelated: I was wondering at grave and measurement, the kanji in the forums is printed as this:
which should be samurai (which also explains the reading し). Looks like the Japanese messed that up while importing.


I admit feeling a little bemused when I learned some radicals that matched kanji after learning the kanji, e.g. ‘Machine’ the kanji is L3 and is apparently ‘pile’ + ‘mouth’ radicals, but then we learn the radical ‘machine’ in L9.

That didn’t make a whole lot of sense to me, but there have only been a couple of times where this has happened.

If I’m honest I rarely use the radicals to remember the kanji so its of little consequence apart from when I really need to differentiate between some that look the same.

Yeah, I’m completely with you on this. Sometimes it just feels completely ad-hoc, like he’s come up with a mnemonic for this specific kanji, without remembering the other six similar kanji he’s about to introduce.

There’s a number of cases (none of which I can specifically call to mind right now) when he’s had to make a mnemonic joining so many diverse sub-units that it comes across like a noodle incident (gee, I think so, Brain, but where are we gonna find a duck and a hose at this hour?), when I’m sitting there going “no, look, if you treat this part like the meaning component that it actually is, then the mnemonic just falls into your lap”.


I don’t really mind the “radical that equals an earlier kanji” thing! It’s a nice indicator of, “okay, we’re going to use this kanji as a building block from now on”. (Though maybe it does feel a little silly to get quizzed on the radicals as if they are “new”… like, I see 下 Below is a level 59 radical; I don’t really think there’s anything to be gained from sending that through the whole SRS process again when the fact that 下 means below is very, very familiar by then.)

But anyway, the inefficiencies bother me more. 台 yields another example: could be ‘machine + heart’ instead of ‘pile + mouth + heart’.

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.