Reading Explanation - same mnemonic different hirigana reading

Is there a way people use to separate the correct hiragana spelling for shared mnemonics?

Here is one I have kind of brute forced over somehow, and is documented already here:
すえ すう
Using in 末 and 数 respectively

こと こた
Using in 事 and 答 respectively

Just can’t get that coat out of apprentice level! Are there any more of these shared readings to expect as I slowly level up?

If you email them with the issue, they might rework the mnemonics.

1 Like

There have been a few attempts at building new sets of “consistent mnemonics” - this thread talks about @sporadic’s attempt, and @Rowena also drops in to discuss her approach (which reminds me a bit of fingerspelling in the sign language context). You might find it an interesting read.

As sporadic notes, there are around 900 unique readings of two morae or less, so this is a pretty big job. You may want to limit this only to readings where, as you mention, there is an ambiguity between the English mnemonic cue and the reading. So perhaps keep 末 as suing Stephen Hawking when you get all the way to the end of the universe and find out it doesn’t actually go on forever as advertised, but replace 数 with すzie Quattro wearing うgg boots while counting a pile of records.


I really should update that thread, now that I’ve had a few months experimenting with the system I evolved.

Basically, it’s as you said, with 2 mora patterns, there are really too many for easily making up enough unique ones and remembering which is which, since none of them come up enough to reinforce.

So I changed to a single mora system similar to what @Rowena came up with. There’s a little less than a hundred of those which is far more tractable. I gave each mora two symbols. First is a character, second an object. Each is chosen for association with the sound, either through their name (Gandhi -> が) or because of associations (The Joker -> は; Santa -> ほ, etc). Second is an object, activity or trait. Many times the character and trait relate to one another (Bigbird and yellow -> き) but other times they’re just each chosen for their association to the sound.

When dealing with a reading, I start with the character for the first mora, then actions for the second and following. If there’s four or more, I’ll split into two characters, but that’s rare for single kanji.

I have some special case mnemonics for very common 2 mora cases, mostly borrowed from WK, like ken the squid, the cyborg, the djinni and the ninja. This is partly because they were so solid since I didn’t use my system until later, and also because they’re so common.

I have two specific mnemonics only for use in vocabulary. One is for gemination and the other for rendaku. I find this is easier than trying to override the default kanji readings. Changing Bumblebee crying like in a Peanuts comic (かわ) to Gandhi crying (がわ) caused too much mental friction (much like the 人 issues everyone has), but it’s much easier just to add Puck (my tricksy little rendaku faerie) to the image and be done.

It isn’t a perfect system. It does take some effort to set up, and sometimes the stories are hard to create. What I have been doing is when doing a lesson, I look at the reading and the WK mnemonic and the character/action combos my system generates, but don’t put effort into an elaborate story until I miss during reviews. Sometimes all it takes is knowing Baby Yoda is involved (よこ. I got very lucky on that one, since they were both picked way before Mandalorian was out) and I’m good to go, without even a story for why he’s involved.

Anyway, probably more than you need to know. It comes in very handy, like around level 10 where WK uses ‘car’ as a mnemonic for like 3 or 4 different kanji, each with different readings. But it’s a lot of work at the outset. Good to have now, though.


I found it was a lot of work at the beginning, too, and still revise the whole lot every week or two to make creating the mnemonics easier (the less frequently-used mora need refreshing), but I have found it works really well as long as I take the time in lessons to create the image.

Love your rendaku-ing Puck - might steal it… :supervillain:


I pretty much do the same, and check to make sure they all still work for me. There have been one or two that I thought would be perfect but just didn’t stick or led me astray, I’ve changed them out.

And to be honest, there are some that are extremely rare in the entire WK database that I just haven’t bothered coming up with a mnemonic yet. I do those as I hit them.

1 Like

I barely use Wanikani’s suggested mnemonics.
I use “Suez Canal” for すえ and “soup” for すう (and “Sudafed” for す—cause it’s just one little capsule)
こと is “kowtow” (not sure how that’s pronounced but it helps me remember regardless), while こた is “quota”


So, to get back to your original question, if it helps, these are my mora based mnemonics for your initial questions:

す - Sue, a little cartoon girl, cute from the front, holding a knife behind her back (think Wednesday Addams, maybe)
え - egged
う - undead

So I’d end up with something like :

  1. 末 Two other girls egged Sue. They’ll come to a bad END.
  2. 数 Sue has raised a zombie horde. There’s so many, it’s hard to COUNT.

Then we have :

こ - Conan
と - towing
た - tanned with sunglasses, little bit skeevy

So we’d have:

  1. 事 - Conan by himself towing a viking ship by a chain. It seems impossible, but his ACTION is making PROGRESS as he strains to pull it along.
  2. 答 - Conan, polished and gleaming from his workout on Muscle Beach, trying to pick up a girl by saying (in bad シュワちゃん/Schwarzenegger voice) “Yo, I am the ANSWER, bay-bee”

Hope that helps see how the system can work.


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