Mnemonics : problems and solutions?

Hi Kimera,

I am a firm believer in the ‘power of the crowd:slight_smile:

I meant limitation as per :

  • number of kanjis I studied ‘deeply’ (*) so far. I’m around 10% versus my objective, so still lots to discover. Gathering others’ input on a larger span will probably speed up my awareness of those problems.
  • areas of difficulties : I may have problems on some areas, while it’s a no-brainer for other people. And vice-versa, others may have problems on something that I did not notice as a potential problem. So having the view of several people will help to identify the

(*) : by ‘deeply’ I mean learning more than the single WK kanji card, but rather a group of several items related to this kanji (ON and KUN readings, writing, a few vocabulary words, sometimes a 四字熟語 using this kanji, etc.)

Actually that’s kind of the idea, at some point I’d like to share back my system to the larger community, rather than just learning the language on my side :slight_smile:

That’s a nice one that will stick with me as well, tx for sharing :slight_smile:
What is great is that you actually created the mnemonics specifically to solve the issues of “じん vs にん”. At least that 's my assumption since you wrote “because you can’t drink gin”. Just curious as this one as been tricky for me : how do you solve the ‘rendaku’ ? Anything you added specifically in the mnemonics or else (rule of thumb, etc.) in order to remember it better ?

For some kanji the wanikani mnemonic makes no sense and is just a word salad, so I come up with my own. Usually, looking at the etymology of the kanji helps a lot. Also somehow coming up with a visual representation for the meaning that’s also visible in the kanji helps a lot. For readings, I usually come up with my own mnemonics because my native language isn’t english, but occasionally I use the wanikani ones too. Sometimes I just can’t remember the mnemonic and I just have to memorize the reading with brute force.


By missing the item about half the time! :slight_smile:

I’m serious! Let the SRS do it’s magic. Same with transitive/intransitive pairs (easier now that they’ve spaced these out intentionally).

If you miss the rendaku, take the time to “hear” the correct pronunciation in your mind’s ear, then move on.

Believe me, if you miss it enough times you’ll eventually see it so often that you’ll never miss it again.

Even works with leeches. Eventually you’ll get so sick of seeing that item that you’ll figure out why you’re missing it and really start focusing on getting it right. Eventually you’ll know it forever.

You’ll start to recognize which words are likely to get the rendaku, but I don’t think it’s worth trying to invent “rules” or tricks. Just trust the SRS.


I can very much second this. Maybe it’s even worth saying the word aloud a couple of times. Usually one of the pronunciations sounds “better” than the rest. Granted, rendaku’d H sounds can be tricky.

I’m afraid you need both: some kind of mnemonics and repetitions.
If you repeat your words often, they will stick for a while. But wait until you start burning them a couple of months later. Chances are, you forgot the word - and if you don’t have any kind of mnemonics to fall back on, you’re lost.

I experienced that when memorizing PIN codes for my banking cards: using the “major system” I associated the numbers with words that where easy to remember. After some time of regular use I knew the pin code and stopped using the associated words every time I used the card. End of story: after not using the card for a while, pin code AND the associated word where gone.

And: you need to detect your Kanji/Vocab in the wild on a regular basis.

1 Like


I’m answering several threads at the same time since it seems to be a best practice (I got an automatic message telling me so when trying to answer to each…)

First, thanks all for the detailed answers, it gives me lots of ideas !
I’ll just comment the various messages to keep the discussion rolling-on.

Great books indeed, and the more recent one “Moonwalking with Einstein” is also a good introduction to those Memory techniques.

This is where I am coming from actually :slight_smile: By adapting such techniques to the study of kanji, it is possible to considerable improve the current system proposed by WK. Eventually WK is a SRS with a content covering 2 dimensions (READING and MEANING). But there is much more left to cover for a kanji on top of reading & meaning…

The best summary ever :grinning:
Let’s be realistic : it is impossible to aim at something that would work ‘for everyone’, because we are all different. What is possible nevertheless, is to add more consistency and logic in order to create a system. The benefits of a more systematic approach being :

  • it allows to cover more than 2 dimensions
  • it makes the recalls much easier / faster

Yes, if there’s one thing where we can all agree on, it’s indeed that there’s no “single magic formula fitting everyone”. All one can do is discover what works best for him/herself, so basically test and adapt !

Yes indeed, creating decent mnemonics is time-consuming… In my case I spend between 30 minutes to 3 hours for each kanji to build my own mnemonics. Not saying 3 hours in a row, it’s actually spread over a few days or even weeks, after reviews or reading some info in various books / websites… or just randomly by running into something in real life that helps for the story I’m struggling with ! I guess ‘creativity’ is not something you can just command with a click :grimacing:

None of the WK members should ever use the adjective ‘lazy’ when referring to himself , we are all the opposite of lazy trying to learn Japanese :wink:
I personally never start by WK, but when I am stuck, I do have a look at various online sources (Koohii, KanjiDamage, WK, RTK, etc.) to search for inspiration.

A question for all : are you creating different stories for each vocabulary words using the same kanji ?
What I’m doing is actually creating a more complex story, about the kanji AND inserting some words using that kanji.
So going more ‘deeply’, learning more than the single WK kanji card, but rather a group of several items related to this kanji : ON and KUN readings, a few vocabulary words, sometimes a 四字熟語 using this kanji, etc.

Yes this is key, and indeed all memory athletes are using repetition to master their own system. The advantage of an initial strong mnemonics is probably that the ‘forgetting curve’ would be softer, but you’ll have to repeat the info anyway to make it stick to long-term memory. Using a SRS being a good way (but not the only one) to optimize the repetition interval.

I recognize this is the magic of WK (both its SRS + its content). You basically just have to sit and do your reviews ! It’s by far the best tool I’ve seen so far. Not perfect, but by following a routine (the one you described, or similar ones by others ), there’s no reason why one couldn’t reach level 60 eventually :grinning:

It’s great to make this conscious effort to analyze an error, because this is basically how we improve our brain. Was listening to an audiobook about that just a few days ago (Mathematical Mindset, by Jo Boaler). Can’t recommend it enough as a parent with young children !


Yes, I usually go about it the other way round as well and that’s why for me the “kanji-only” stage of each level is the most troublesome. I usually try to incorporate into the learning process words which use a given kanji to nail down the sentiment behind a kanji, rather than its literal word-to-word translation/meaning, because long-term that’s not very relevant I think. Depending on how much effort I want to put into a kanji I would distinguish between:
0 (no effort) - 3-5 reps of a kanji with maybe 2 short breaks in-between (long enough to focus on something else and get distracted from the process of actively learning a kanji)
5 (extra effort) - put additional words using the kanji with the reading I’m struggling with into my Anki decks
10 (lotsa effort) - try to incorporate the kanji into an already established phonetic net of other kanji, with extra focus for reinforcement, etc.

I don’t ever, however, create stories for vocab items. These I usually figure out from the kanji and only some oddballs require extra mental leg work :slight_smile: .


What a terrific response! Thanks for taking the time.

Oh yeah, that sounds familiar. (laugh)

The last several review sessions on a level are by far the most fun: you only get vocabulary lessons and most of the kanji have already started to sink in. Conversely, I sorta dread the first few days of a new level knowing I’ve got a week or so of unfamiliar “pink” to deal with.

I have an advantage over many here in that I’ve been speaking Japanese for decades (albeit poorly). I already had a reasonably broad vocabulary. Trust me: learning the kanji in vocabulary words I already “knew” noticeably improves my conversational Japanese.

Like you, I also don’t usually bother with mnemonics for vocabulary (often not even bothering to read the provided ones). Maybe 60% to 80% of the time I already know the word. The kanji in the others usually give me enough of a hint regarding the meaning, so the main things to memorize are pronunciation, rendaku, and transitive/intransitive pairs. As we’ve already discussed I mostly just rely on repetition for those and don’t worry about mnemonics unless I run across something particularly intransigent (usually a leech).

For radicals and kanji, though, I mostly depend on the provided mnemonics for the first few reviews.

One thing that used to bug me frequently was when I mistakenly filed a character into a subtly incorrect mental filing bin. Fairly often I’d come across a kanji (or even vocabulary) that I didn’t realize was for a word I already “knew.” Instead of filing it away with the word/concept I already knew, I’d create a new mental entry. It took me an embarrassingly long time to realize that the verb 存じる (ぞんじる — to know) was the same root of the polite phrase “ご存知ですか?” (“Do you know …?” or “Are you familiar with …”?).

Similar things happened when I got the nuance wrong (e.g. a thing’s “use”/utility is very different than “use” as a verb, and Japanese has several different words in each category).

The new “context” section, giving you a preview of upcoming vocabulary has been amazingly helpful with this. I’m so happy that the pedagogical geniuses of WK added this!


Hahahaha, very relatable!

Oh yes, I very often have this problem when I first learn a kanji through multiple vocab items I put into my Anki sets and then later it appears on WaniKani and it takes me an embarrassingly long while to realize I actually already “know” this kanji. That’s not always the case, but often enough.

Funny that you mention 存じる - for me this one and 存在 that came up in anime recently (which obviously appeared on WaniKani ages before, but I failed to make the connection :man_facepalming: ) helped me to distinguish 存 from 在. 存じる with ご存知ですか I somehow connected, because the core phonetics overlap, but it’s enough for a word to be pronounced slightly differently in a show, news report or a song and I’m helpless :sweat_smile: .

Oh yes, this is especially tricky. I’m just now slowly learning the various nuances of the “life” words (生活、命、人生, etc.), but today for instance I was doing some research on “manufacturing” and realized I don’t understand the nuance between the various terms. Here’s the article by the way: 「生産」「製造」「製作」の違い | 社会人の教科書


I make up my own mnemonics and I have a system!!

I generally use the radicals and some of the reading mnemonics that WK uses except some of the old ones have stuck so instead of “knife” I still use “ribs” and instead of “sheep” I still use “Charlie Sheen”.

But I have a few rules. The reading mnemonics must be standardized. So if it’s “きゅう” it’s ALWAYS going to be “Cutie Honey”. Not Cutie Honey for the 級 and then “Cucumber” or something for 休.
The other huge thing that’s helpful for me is that the reading mnemonics for me are always people or places. And my little mnemonic sentences always have the same order RADICALS → READING → MEANING Because that’s how the brain process for me works, see the kanji, look at the radicals, that triggers me to recall the reading, then that triggers me to recall the meaning.

Example for 坊:
RADICALS- 土 dirtdirection
READING- ボウ bowling alley

If you follow the direction of the dirt, it’ll bring you to the bowling alley where you can find your MONK.


Interesting! WK itself appears to follow the first rule.

It’s similar to the approach memory athletes use to memorize standardized but conceptual information like numbers, playing card values/suits, etc. You create a system to associate highly visual and memorable (but composable) “tags” to the otherwise opaque information.

For example, in the system I learned, the numeral 1 gets the “T” or “D” sound (dakuten!), and the digit 5 gets the “L” sound. To use the system you must first rote memorize the sounds for all 10 digits. If you wanted to memorize something involving the number 15, for example, you might incorporate the word “TaiL” or “TaLL” in your absurd, perverse, or embarrassing mental story. The number 155 might use “TaiL Law”.

I’m not sure how many standardized reading mnemonics WK uses (KOU-ichi, GENji, Mrs. CHOU, little KYOuto, big KYOUto, KYUUcumber, etc.) but I doubt it’s more than a couple of hundred items at most.

It might even be possible to devise an even more decomposable system for any arbitrary reading (say, one term for each of the 50 五十音 with maybe some system for dakuten/handakuten/sokuon).

I doubt it’s worth the effort though. For some reason, I personally find it easier to memorize readings, even without mnemonics, than to memorize meanings.

The number system I use doesn’t distinguish between “tail,” “tall,” or even “ToweL” though: any of those could be used for 15. So I’m not sure that it’s really a problem to use different words to memorize the same kanji reading (“cutie honey” / “cucumber”) — either should help you recall the “kyuu” sound which is the whole point. Both you and WK seem to feel there’s an advantage to always using the same reading mnemonic, but I’m not sure why it’s an advantage. I guess it does make it a little easier/quicker to devise a mnemonic story.

Also, doesn’t your RADICALS → READING → MEANING system break down since almost all kanji have multiple readings? WK understandably has you memorize just one reading at first, but I don’t think that particular reading is any more important to memorize than any other. It seems WK’s system of separate mnemonic stories for meanings and each reading (kanji + vocabulary items) might work better.

Unrelated, but I’m curious about WK’s system for deciding which reading to teach first. It’s usually, but not always, 音読み. I’m sure there is some sort of rationale, but I don’t know what it is. It probably has something to do with usage frequency.

1 Like

Other than the above, I’ve seen JORm (じょう) and TOUKYOU (used as a whole or split into とう and きょう, depending on the WK item). Can’t remember the others.

I usually derive the meanings from the kanji themselves, but that then sidesteps the phonetic component which might become disjointed, while WaniKani tries to keep things together in their mnemonics.

As I wrote in another thread, for me it’s easier to aggregate around phonemes and couple them with general meanings:

  • ちょう (long, chief, significant (?))
  • しょく (food related)
  • しょう (correct, proper)

The rest is built out from that so instead of:
abstract concept/persona ->> phoneme ->> story
I get:
phoneme ->> sentiment ->> story? (usually a fairly short quasi-visual connection)

I also ignore this process when it’s possible to derive the reading of a kanji from one of the elements. So for instance:

I would just do:
坊 ->> contains 方 ->> has the usual readings ほう or ぼう (more often ぼう when it’s a radical)

But doesn’t that generate too many stories with possible overlaps and as @YannickFrance (or someone else, can’t remember now, sorry) mentioned, lead to interference?

I think it’s backed up by the vocab they teach. So kind of assuming vocab usefulness is somewhat a priority, the available vocab items point to either 訓読み or 音読み. The reason I think that’s the case is because sometimes the meaning of the kanji is also biased towards a specific direction which is then noticeable in the vocab items.

1 Like

This is I think one of their biggest missteps, translating 台 as “machine” instead of “platform”, because it really only means machine in the context of being the counter for large machines; in nearly every compound it means “platform”. It’s much easier to remember 舞台 → “dance platform” → “stage” than some weird mnemonic about a “dance machine” having something to do with a stage.


Yup, exactly. While this may require some major changes to WaniKani, I thought of the pros and cons and reported this, because using “machine” for 台 goes beyond funny radical names and is genuinely misleading to current and new learners. They’re working on it :slight_smile: .

I don’t mind alligators, mohawks and mona lisas, because while weird, these can be useful for memorizing radicals and kanji.

For some of the remaining items it’s usually enough to add a user synonym :slight_smile: .

1 Like

Huh. That one catches me by surprise.

To me, 台 on its lonesome most definitely means “counter for machines/systems/vehicles/etc.” Probably because I worked in Japan selling “systems” for so long.

You’re both absolutely right that it also means stand/rack/table/bench/podium/platform/stage, especially in compound vocabulary words, but my brain really doesn’t think of that character that way when I see it on its own for some reason (though I know both meanings).

To me, the 一台、二台、meaning is more “core” to that character than the 舞台, 台所、踏み台, etc. meaning.

I’m likely way off base, but to me, the “machine” meaning is the core kanji meaning, and the “platform/table/stand” meaning is for vocabulary.

For what it’s worth, dictionaries seem to back up your interpretation more than mine (“platform” as the primary meaning). The Outlier dictionary says the original meaning of the character was a “grammatical, more abstract version of ‘to take.’” Don’t ask me how you get from there to “machine” or “platform”!

I could never be a linguist. Words with multiple meanings are so confusing. English is even worse!

Anyway, I’m getting way off-topic: none of this has much to do with mnemonics.

Regardless, I think we all agree that WK alone will never suffice to learn Japanese. WK does a great job of helping people memorize several thousand kanji and vocabulary items quickly, but you simply must supplement WK with other tools (dictionaries, grammar training, speaking/comprehension training, readers, etc.) to become fluent.

That’s not a knock on WK!

In my opinion, adding more functionality (complexity) would detract greatly from its “teach kanji quickly” mission. The same holds true if they added more “ratholes” for further research. I think it’s best to memorize thousands of kanji and vocabulary as early as possible in your studies, even if you only learn a subset of the meanings. It will better prepare you to explore all the other nuances and complexities as you continue your studies.


The ultimate version of WaniKani, albeit one that would likely take much longer to complete, would be one that lets you make your own mnemonics for every kanji. You just have to memorise the radicals, and from there you construct your own story.

1 Like

Even better if it gets you to floss regularly and pay your bills at 8:00 a.m. on the 1st of every month. :wink:

Having tried both WK and making my own Anki decks, I’ve conclusively proven that spoon feeding and a great UI works better for me, at least.

I agree that forcing us to make up our own stories would probably ”work” better, but I prefer the lazy route most of the time. I think forcing personal stories 100% of the time would probably increase retention but at the cost of more drop-outs.


I highly suggest you read my post “Hump Day: Level 40 & Beyond” and then from there the post followup which is about Mentalists and Memory Palaces. Good luck!

Here’s the link.

Yup. Moonwalking with Einstein was brought up earlier (written well after I read Harry Lorayne’s memory books).

Your examples in that thread explain why I’ve found the Niai visually similar kanji user script so helpful. I often have to miss a review a few times before I realize I’m confusing it with a different character. Usually it’s already present in the list provided by the script, but the ability to add your own characters with that script has proven to be a godsend. I’ll add whatever character I’m conflating it with as well as a note for whatever memory trick I use to distinguish them.