こんにちは、みなさん！
I’m building a small Japanese-learning app called Bento Cards. The idea came from struggling with “leech” words, the vocabulary that keeps coming back in reviews because it just doesn’t stick.
I wanted a more visual way to remember those difficult words, so Bento Cards uses mnemonic cards to make vocabulary feel more memorable.
It’s still early, and I’d love feedback from WaniKani users / Japanese learners because this community has a lot of experience with mnemonics and SRS-style learning.
I’m especially curious about:
-
whether the mnemonic cards feel useful or memorable
-
whether the visual style helps or distracts
-
what information a good vocab card should include
I’m not trying to make a sales post, I’m mainly looking for honest feedback so I can improve it.
Link: http://bentocards.app
よろしくお願いします 
IMHO it’s at least potentially useful to me, although I do have some doubts.
A lot will depend on how ‘good’ the written mnemonics and illustrations turn out to be.
Is it mainly geared towards nouns? Is it intended for vocab only, or also for kanji and radicals and phrases and verbs, etc?
A lot of my leeches are not simple nouns, for example.
And a lot of leeches involve remembering factors such as transitive vs. intransitive, or remembering the exact wording that WK is looking for (such as forgetting to add the word ‘something’ to the meaning) - and such examples are likely to be difficult to convey in the type of card that you’re showing.
Also, I find that a lot of the WK mnemonics are less than helpful to me, and the few examples of yours that I’ve seen do not particulary encourage me either - frankly the spinach hen mnemonic seemed pretty lame to me.
Forging too strong a link between mnemonics and the vocab, whether those mnemonics are word-based or visual, may actually turn out to be detrimental to learning.
It also occurs to me that a ‘suitable for intended task’ test would be necessary.
For example, presenting a kanji or word and then recalling meaning is a different task than recalling the reading - are you trying to handle both of those within a single card? That may be problematic.
I’m also unsure of whether or not you may be attempting to address some of the bases that WK ignores, such as remembering the word or kanji when presented with the English definition, or dealing with context sentences beyond just presenting them, or other WK blind spots. But maybe those things wouldn’t translate well into the visual card paradigm.
Also, some leeches may be due to confusion with visually similar kanji, which may not fit neatly into your current framework or approach.
Also wondering about complicating factors such as onyomi vs. kunyomi, or handling ateji with mnemonics, etc.
Yeah, if your goal is to build a truly useful learning tool as opposed to a ‘feels good at first glance but doesn’t really deliver on the promise’ toy app, it may require a lot more depth.
BTW, are you relying on AI to generate these? If so, lots of potential pitfalls come to mind.
Maybe making user interactivity and modifications may help - perhaps you’re already doing that or contemplating it.
Thank you for the honest feedback, it really means a lot.
You’re right that many of the hardest leeches aren’t simple nouns. Transitive/intransitive pairs, visually similar kanji, subtle meaning differences, readings vs meanings, context-dependent words, and even WK answer precision are all much harder problems than “remember this word.”
At the moment Bento Cards is mainly aimed at beginners and is closer to a proof of concept than a complete learning system. The current mnemonics mostly rely on pronunciation and meaning associations because I wanted to keep the first version simple and see whether the idea itself has value.
I completely agree that if the mnemonic isn’t good enough, it can become distracting rather than helpful. That’s actually one of the things I’m trying to learn from people’s feedback.
The cards are AI-generated, and right now the main control is simply regenerating them until you find one that clicks. In the future I’d like to give users much more control, for example:
I also like your point about user modification. My own experience is that the best mnemonics are often personal, so being able to edit or guide the generation is something I’ve been thinking about.
For now I’m mostly trying to answer a simpler question: can visual mnemonic cards help with at least some of the words that refuse to stick? If the answer is yes, then I can start tackling the harder cases you mentioned.
Thanks again for taking the time to write such a thoughtful response. These are exactly the kinds of challenges I need to think about if I want this to become a genuinely useful learning tool.