Mnemonic improvements PROPOSAL! ✨

TL;DR - we want to make some mnemonic changes for characters and want to know your thoughts on it before proceeding.

A common point of feedback we have gotten is that our current mnemonics are a bit hard to process.

Current mnemonic approach

To illustrate, let’s go over the mnemonic for 敵 from our traditional course. It is pronounced dí and means “enemy”.

[di-] Diddy Kong is [2] inside the living room of the [-_] house, trying to dodge the crazy 攵 foreman’s attacks. The 攵 foreman, his 敵 enemy, keeps on throwing 啇 antiques at him, and one hits him in the tail. The foreman screams that man and monkey are 敵 enemies, and throws his yellow construction hat at him too.

The story itself is not bad! However, there are some things that could be better.

  1. The meaning and pronunciation is jumbled together within a single story, making picking out each individually a bit more mental effort. It also makes it a bit longer.
  2. The story begins with pronunciation elements, but should probably begin with components. After all, when we look at a character, we start with the components and use that to recall the story!
  3. Besides the phrase “living room of house”, there are not many clues that this is indeed the living room. It should include something mentioning a carpet, or a couch, or a TV - anything that conveys that feeling of location.

Proposed improvements

To address these issues, we are thinking of doing the following:

  1. Split the mnemonic into a “meaning” mnemonic and a “pronunciation” mnemonic.
  2. Ensure that the “meaning” mnemonic places the component names earlier on, and the meaning later on in the story.
  3. Ensure that there are more subtle “location hints” within the mnemonic.
  4. Redesign the lesson flow to better fit a split mnemonic.

Here is an example of a split mnemonic for that same character:

Meaning mnemonic

“Don’t throw the 啇 antique at the 攵 foreman! Doing so will just make him your life-long 敵 enemy.”

It begins with the components and leads to the meaning. No pronunciation information is included, and we try to make it as brief as possible to aid with memory.

Pronunciation mnemonic

“The 敵 enemy of the foreman, [di-] Diddy Kong, hangs from the ceiling fan of [2] inside the living room of the [-_] house, tossing antiques at him.”

It begins with the meaning and weaves in the pronunciation elements in a more naturally worded manner. It will include some components as well, but they won’t be highlighted and will be omitted for the sake of brevity as appropriate.

Additional new mnemonic thoughts

The amount of information and the story itself did not change much, it is just split up to make it more digestible.

The two split mnemonics, as we see, are related, allowing them to be remembered individually or in tandem, depending on the needs and learning style of the user.

The process of recall would generally be:

look at components → recall meaning via mnemonic → recall pronunciation via closely related mnemonic

Let us know your thoughts!!

What do you think? Is this a clear improvement over what we had before?

We already have some 3,000+ character mnemonics, so updating them is quite the effort. However, we are happy to do so if there is a clear benefit in doing so. Our goal is to be the best platform for learning Chinese characters, after all! We ourselves quite like this new approach, but want to know that this is something our users want before proceeding.

What do you think? Anything else we could do to make our mnemonics even better? Of course, our mnemonics will never be able to satisfy everyone, but we are trying to find a path that is most suitable to the majority of our users.


This looks great to me and is a big improvement on the existing :+1:

Presumably the pronunciation card comes up some designated time after the meaning card?

One thought, but it would require quite a bit of upfront work (which might be too much for you guys atm but it could potentially save you a lot of backend work in other regards) but an ability to make custom mnemonics “public” with a checkbox or the like, and then for a user (perhaps via a “public mnemonic” dropdown) can view the ten most upvoted mnemonics within a given card and apply it to their accounts card (hopefully that made sense) - This way if the default mnemonic doesn’t fit a given style then we can quickly grab one which hopefully does from another user. Plus this method would give you guys some valuable data :slight_smile:

Next one would be more down to server costs your end I’d imagine, but having an image entry/upload for image mnemonics would be good too, much like the radicals aid in association, and then the audio does, images are another “free” form of reinforcment and association


I love the idea of splitting up the meaning mnemonics and pronunciation mnemonics. I think it will benefit users and will be worth the effort. It was easier to follow along with your example split up like that. Anything to make it easier, and thus faster, on the brain!

1 Like

Going from the components to the pronunciation and meaning is a good idea to make the mnemonics easier to recall. What I’m not so sure about is whether it’s better splitting it into two different parts. As I see it, one mnemonic implies having to remember one thing, while having it split into two makes me feel as if I had to recall two independent pieces of information (even if the two mnemonics were closely related, it’d be portrayed like that). Besides, the pronunciation and the meaning are both essential parts of each character, so why not keep it together? Is it not going to still be necessary to read the whole mnemonic always anyway?

But if the mnemonic is so complex (particularly on more complex characters) to the point where you might not want to use it in the first place, then this doesn’t work. Breaking things into two at least makes things somewhat more manageable. particularly if the meaning mnemonic doesn’t click but the reading one does, least then your not wasting the whole thing.

I see your point, but you are still remembering two core things (reading and meaning) even if it’s in one mnemonic, which is just harder to digest. 50% of the mnemonic info would be wasted when the card comes up for review in each form (I hope this makes sense; you are wasting 50% more brainpower going over info in your head that is not relevant to the meaning or vice versa, the reading, respectively). Also, what I’ve found is that later on, once you become more familiar with the character in question, you instinctively fill out the reading without even hunting for the English meaning (this is a good sign), and it also means you are not going to self-penalise in the event you do actually forget the meaning (for example) but still recall the reading.

My assumption is that the reading (or pronunciation, if you prefer) card is shown +10 (or X number) cards after the meaning (at which point the meaning is in short- to mid-term memory), and then when the reading card is shown, the meaning association is “free” and requires no additional effort on the user’s part. All that is to say, yeah, you are going to get all the info eventually, but it’s broken down, doesn’t overwhelm, and is easier to remember as a whole.

Also the meaning mnemonic and reading mnemonic in their split form will (presumably) be shown in both the reading card and the meaning card at the same time in both cards, so I’m kind of failing to see the issue from your perspective. Just read them back to back if that fits your learning style, no?

Also the other elephant in the room. Hanzihero is essentially the Chinese equivalent of Wanikani, even if that’s not the intention, it will be the perception. A lot of users will be making their way here from Wanikani; as such, the expectations will be that the basics are the same between the two (split mnemonics, amongst other things).

Anyway, all just imo, each to their own and all that :slight_smile:

1 Like

This looks amazing!! To me the mnemonic in this example goes from being very difficult to parse out in the current form to being very digestible and easy to remember.

Thank you so much for taking our suggestions into account :pray:


I see what you mean regarding the “waste” of brainpower when doing the flashcards. My perspective is that you’re going to have to remember both the meaning and the pronunciation when you encounter the characters in the wild, so having one mnemonic fits that purpose. I don’t see it as a negative practicing recalling both elements at the same time because that’s what I have to do later, if that makes sense.

I’m not sure what you meant here. What self-penalisation is there with one mnemonic?

I thought the meaning and pronunciation would still be learned at the same time, so now I see what you are going for. I understand splitting the mnemonic is useful in this case because the first part has become useless by the time the user gets to the second part, so it’s superfluous information. What I wonder then is how the number of reviews is going to be adjusted. If each learning element is just the pronunciation or the meaning, the number of learning points per day is effectively halved.

If what I wrote earlier is true, the learning style of reading both mnemonics wouldn’t be possible because of the way they’re split in the reviews. By the time I get to the second part of the mnemonic, I should’ve already learned the first part, so that I don’t need the first mnemonic. And reading both mnemonics from the get-go wouldn’t be effective if I’m only going to review one part in the near future. Whether the mnemonic is one big or two smaller paragraphs obviously makes little difference; the order in which the new reviews are shown does matter, however, if I understood the system you described correctly.
All in all, I think it’s just a question of taste and maybe reworking all the mnemonics and the review system in that way is a lot of work for little actual change in the results of the learning process. Anyway, everyone has a different point of view, as you said, and I also can’t really know how much effort it’d entail, so this is just my two cents.

1 Like

Sorry, I didn’t do a great job of explaining this point. Basically (this is similar to what I touched on in another topic) You often hear from people who are fluent in another language that they have trouble translating from the fluent to the native on the fly as the two “occupy different parts of the brain” , this isn’t because they don’t know the reading and meaning relation but (in the case of Chinese) their brain is in Chinese mode (or, whatever language) and trying to then recall the english meaning of a well known word (whether that be via a mnemonic or otherwise) direcly before or after can throw a spanner in the works, (though just to note, I’m not saying doing that in a mnemonic is a bad thing initially, I actually think it’s a very good thing). For example, if I’m doing my Japanese, and I’m presented with the character 猫 my brain is in Japanese mode, I see the animal in my head and know the reading is “Neko” because I’ve went over this card so many times, I know it means “cat” but my brain isn’t in english mode and I don’t even consider that side of the equation (at this point). I think this is good thing and shows language progression and the split approach kind of fits this. The inverse, I have to remember both (or it’s required in the mnemonic to get there, either for the meaning / reading) then I think discourages this and I think I would “self penalise” which brings in my mind negative associations.

Hopefully that makes sense, I’m terrible at explaining things :upside_down_face: :smiley:

Well it would only be singular Hanzi characters at that point right? So I’m not sure it would be quite as much as half (?)

Words, components and such wouldn’t get affected by it, so it’s true that the overall learning speed is not halved. The one concerning the characters would slow down significantly, however, and I think that issue should be addressed if they decide to implement that split system, since the point of reworking the mnemonics is getting to learn at the same speed (or faster even) with less effort. If I wanted to slow down the amount of elements I learn per day, I can already do that.

Regarding this, by the way, there’s also the issue about how word mnemonics are going to work. Many only need the meanig, since the pronunciation doesn’t change, but it does sometimes, so what is the approach in that case? If a change in pronunciation means a split mnemonic as well, what I said about the single characters also applies here.

Thanks for the feedback, everyone.

How will character lessons/reviews will work, now?

Characters will show up in your reviews and lessons the same way as they always have. We’ll only tweak the lesson carousel a bit to better fit the split mnemonics. Today we have the following tabs in the lesson carousel:

  • Components - point out the components in the character
  • Sounds - point out the sounds in the character
  • Mnemonic - the single mnemonic
  • Words - some example words that the character will have

The new character lesson carousel will have these tabs:

  • Components - point out the components in the character
  • Meaning - meaning + meaning mnemonic + alternate meanings + user-editable notes
  • Sounds - point out the sounds in the character
  • Pronunciation - pronunciation + pronunciation mnemonic + user-editable notes
  • Words - some example words that the character will have.

How will vocabulary words look?

They will look and behave mostly the same. We do have a plan to tweak the lesson carousel for them a bit, but it is mostly unrelated to the changes this thread is about. They’ll continue to have a single mnemonic for now.

Components → pronuncation/meaning is generally easier. (x3)

I’m glad that everyone seems to be in agreement about this point!

Hesitation about splitting the mnemonic.

I totally hear you on this one, as I initially was of the same thought. However, after trying this internally for a week or so here, we have found that the split ones are almost universally easier to remember. As noted, though they are split into two, they still mostly form a single cohesive story in a sense, so that benefit remains.

I hope that address most points. There was a lot of discussion, so feel free to reply to me if there are any specific questions or concerns I did not address.


This would be amazing