[Userscript] WaniKani Kanji Review Vocabulary List

WaniKani API And Third-Party Apps
21

Although it’s not quite the same thing, be sure to check out the ConfusionGuesser userscript if you haven’t seen it. This one activates upon entering the wrong answer, and I believe it can show visually similar kanji that you might have been thinking of when answering. I mostly do my reviews on the Flaming Durtles app (except for kanji now!), but any time I do reviews through the site, I always appreciate the ConfusionGuesser popping up when I make a mistake.

This one would be outside of my scope to implement because I’ve already learned them. I’d have no way to measure if it’s helping any.

My path to learning these was:

  1. Read lots, so I get general exposure to them. It didn’t help with distinguishing them, but I got a very general feel for their usage. (It’s possible that encountering many verb pairs in WaniKani lessons will have a similar effect to get generally used to them.)

  2. Watch Cure Dolly’s transitivity video. Since I was generally familiar with the verbs, watching this video made distinguishing many (not all) of them unforgettable for me.

Time to brush up on Anki or Kitsun and making sentence cards :wink:

For example, I always get the meaning for 催促 wrong (confuse it with another word, since there as many similar-meaning words learned back to back). So I plan to make an Anki card with the sentence and a screenshot of the video from this Twitter post. Hopefully that will help me properly learn the meaning.

2 Likes
22

already using the confusion guesser script… sometimes it helps but then when it comes up again oh yeah that one I screwed it up again…my memory is horrible…

as far as the self/other move pairs…just meant it’s a nice to have…suggestion or possible thing that could be implemented in a script like this…I have tons problems to be or not to be FFS can’t remember and even watching the cure dolly video about 8 times over the past few years I still often forget which it is and then the ones that flip drive me mad … There’s one vocab recently it’s new level 31 but I’ve reviewed it 51 times and it’s still apprentice because I finally learn it’s to be…then it gets to guru2 and because that interval is too dang long for me…i screw it up and eventually it works it way around again…failed it again a few min ago…just don’t have the memory of a young person anymore

Don’t want to use anki and all the other srs on top of what I’m already doing…already overloaded with SRS and it sucks any fun out of learning… doing WK, bunpro and then adding anki on top…it’s just exhausting… (for me personally). I have tried many times to go back to anki and do vocab there but in the end it’s just a miserable experience for me.

I also use flaming durtles a lot for reviews…so hopefully there will be more kanji tonight … will see how this script goes … only made these mentions because I could see quite a bit of potential in using something like this w/o adding an external SRS.

1 Like
23

I’ll admit, this would probably happen for me if I were doing 100% of my reviews on the site. (I’m still doing radical and vocabulary reviews on Flaming Durtles. I’m grateful that it has an advanced review setting to order reviews by stage, so I can do all radical reviews, then another advanced review setting to reverse review order so I can do all the vocabulary reviews. That leaves kanji to review on the site.)

Well that gives me something to look forward to.

It could partly be a matter of how you’re going about making the cards/notes. I know I went through a lot of things that didn’t work. But the number of daily reviews you have total can impact things as well. For example, the only way I’ve been able to take on Anki for daily reviews was to find ways to reduce my WaniKani daily reviews (including stopping lessons for w while). So, I definitely empathize with you on this one!

By the way, I don’t mean to sound dismissive of your thoughts and ideas. If I can salt these kanji leeches of mine, I will definitely be doubling down on finding ways to take on the vocabulary leeches.

1 Like
24
don't think you're being dismissive at all...no worries ... all constructive discussion :)

for me personally too many SRS just wipes all motivation away…and no matter what I ended up doing just seemed like all I had time for was SRS and no actual fun things…real motivation killer - same reason I gave up on KW

can you imagine having to give up all the things you read because you are overwhelmed by SRS :wink: the horror hehe

when I finally got to see the script in action…was thinking … this type of script and the layout has some real great possibilities and wanted to share some random ideas… (good, meh, and maybe horrible ideas) but wanted to throw them out.

really pleased you chose to develop this and share this script....really hoping this one works to kill those evil leeches....

ultimately the best thing would be WK to finally actually take action and build something proper instead of relying on users to try to script something that should really be a default feature in any SRS (gonna keep calling out WK on this until they do…maybe they never will but it’s annoying - looking at 2.5 to 3 years to finish WK at the rate I’m forced to go because of the leechy hell)

1 Like
25

I think personally I’d prefer if the script was more targeted towards leeches. It should be possible to identify leeches rather easily, and then under those you could put a random set of 3-4 visually similar kanji and vocabs, much like the Leech Trainer script. This would make the distinction between similar words clearer, while also avoiding the over-exposure issue from adding too much context to items you have no problem with.

I’m tempted to write one for myself now, see how this goes. Thanks for sharing! You gave me lots of good ideas

2 Likes
26

The main limitation at this time is that I wrote the userscript without it accessing any user data. I’d definitely like to have it have some sort of leech-detection logic (even if it’s just for kanji that were unlocked more than X days ago, where X is a user-configurable option).

I currently have the “luxury” of not having done any kanji lessons in recent months (aside from 苺, which I already knew), so I don’t have any recent kanji in rotation at this time. It’s definitely something I’ll likely later want to update this script for.

2 Likes
27

FIFY

@rfindley have you seen anything to support the idea that typing helps retention? I’m only noticing a minor loss of accuracy but am not convinced this is because I switched to Anki mode. It oculd be a contributing factor, but I think that it’s more likely to be that I do around 20 reviews per minute in Anki mode. To me, the minor loss of accuracy is definitely acceptable given the staggering amount of time saved by using Anki Mode, but I’m still wondering whether there is any evidence at all to support WK’s implication that typing increases retention.

1 Like
28

Anecdotally, yes… I’ve seen several users report a negative impact from switching to Anki mode. I’m not sure to what extent, and I suspect it would make a difference what you’re thinking as you answer.

If I had to speculate at the neurological cause:

  1. If the time you spend on an item is so short that you’re not doing anything but recall before quickly moving on, then you may not be giving your brain enough time to strengthen the revisited connections. I’ve personally noticed that phenomenon before on non-Japanese studies in the past.
  2. Any additional portions of your brain – such as motor skills, verbal, visual, or thought-association – gives your neurons more inputs with which to discover and encode associations, which gives you more potential avenues for recall. Specifically, typing (if you have good typing skills, anyway) would likely be encoded in the cerebellum, which plays a role in skill- and sequence-related memory.
3 Likes
29

Interesting. Here are some thoughts, numbered to mirror your list:

  1. Could you clarify what you mean by revisited connections? Also, if you’re able to recall on a split second, is this not also a good sign?

  2. So if someone like me, who is terrible and slow at typing, uses in-context reading for “additional avenues” of recall, would this non-productive avenue be of much use in your opinion? After all, it is no more than additional exposure outside of SRS but within a wider context.

1 Like
30

[…typing on tablet is so slow!]

  1. ‘Revisited’ simply meaning that you are activating those neurons during recall. Being able to recall quickly does mean that you’ve created an ideally short path to recall. And, indeed, any activation of a neuron does strengthen the synapses that triggered the firing of the neuron. But there’s also a mechanism where information sort of ‘resonates’ through the brain, triggering additional recalled associations (e.g. things that the item reminds you of). That resonance takes a measurable amount of time, and if you move on before the propagation makes it back to the original set of neurons, then you are missing an opportunity to strengthen those secondary connections. It’s sort of like this: Suppose you look at an old school yearbook and see a photo of someone, and you recall their name. If you immediately move on to the next photo, then you are only strengthening the association of photo and name. But if you take a moment to let some memories of that person surface, then you are also strengthening those additional connections, which strengthens your overall ability to recall the person.
  2. Yes, ultimately all context is likely to help.
2 Likes
31

Thanks for making this! ^>^ I think the concept is great and I think it will help a lot when figuring out some of those kanji that I frankly know, but I still manage to get wrong in certain reviews…:sweat:

1 Like
32

What would be the “extra connections” in the case of a Kanji vocab assuming that the entire universe of connections consists of connections between WK content? Would it be the radicals and mnemonics?
For vocab, would it be the individual Kanji and mnemonics and even further down the line the radicals that make up the Kanji?

1 Like
33

And this brings me to a thought I had when reading your prior comment.

Compared with Anki mode, I wonder how much each of the following impacts association:

  • Touch-key typing on a physical keyboard.
  • Hunt-n-peck typing on a physical keyboard.
  • Hunt-n-peck typing on a phone or tablet.
  • Whatever people do to type faster on a phone or tablet. (I’m not one of them, so I don’t know how to define it.)

Like, if I spend a lot of time slowly keying in letters on smartphone, I wonder how beneficial that is (or isn’t) versus Anki style.

I decided at one point that I valued reducing review time to increase available reading time, which has been a quite satisfactory trade-off for me. (I can type much more quickly on a physical keyboard, but the majority of my SRS time is away from a keyboard.)

34

Well, the “entire universe of connections” would also include what’s in your head apart from Wanikani, such as any pre-existing context or notions of a word. Heck, even looking at 馬 and thinking that the four strokes at the bottom remind you of a horse’s legs. And yes, most mnemonics are actually a way of linking various pre-existing concepts in your head to a Japanese word or kanji.

I’m not suggesting that you deliberately stop to think about the mnemonics, though. I’ve simply noticed in prior experiences that you can get into a sort of ‘dismissive’ state of mind while drilling information too quickly (Koichi has a term he uses for the same phenomenon, I think). When I noticed what was happening (I felt like I wasn’t improving my memorization after several rounds of drilling, and I wondered why), I slowed down to allow a moment for the information to ‘soak in’ for a few seconds before moving on. There’s something about that few seconds that got me out of that dismissive state, and memorization immediately improved. (Sorry, it’s hard to describe, but I could feel the difference. I suspect that having a lot of neuroscience background helps in noticing subtle aspects of what’s going on in the brain. And curiousity makes you more observant, I suppose). Anyway, it seemed reasonable that the resonance is contributing to that.

Yes, that’s a great trade-off! Studying kanji and vocab in isolation is good for quickly ramping up your initial knowledge, but reading is what’s going to really glue all of that information together and help it truly become integrated.

2 Likes
35

so finally got to use this script a bit more doing all the reviews on the computer instead of the phone … found some interesting things (everything works no need to be dismissive of my thoughts now … lol j/k)

worked really really well for this one…read the first 3 vocab w/o even thinking and went yeah I know this but looked at the kanji and went…hmm…w/o that vocab don’t think I could have gotten it…would have probably guessed the reading correctly but wouldn’t have been certain.

image

a place where it is 100% the computer/script/programmers fault…yeah not the ID-10-T error or anything…is meaning

image

really hate these types of things because I end up borrow, lend, rent…oh FFS which one is it… and then all I do is mix up synonyms or ideas and have yet to figure out how to fix this (just my lousy memory really) [in this case it’s interesting, not sure if the script will make this problem worse by seeing the vocab or not - time will tell] … in general, I try not to add synonyms to kanji as really it’s more a matter of mixing up vocab / kanji meanings and intermixing ideas and I get all turned around…hence…just make more leeches (seems like the harder I try the worse it gets…anyway)

image
This was was interesting because the first vocab is a leech…couldn’t read it right off…but it did help with the confusion, (for some stupid reason mix this up with another kanji)…the vocab helped. That first vocab keep forgetting was a nice little mini leech review sort of built into the kanji review. Admittedly it took me about 3 min to recall though so not great but at least maybe that will help it stick when the vocab comes around again.

image

Your script saved me on this one…my first thought was arakajime…no wait a freaking minute…that’s not right given the vocab… interestingly though only really knew one vocab so had to think about this one a bit…wasn’t confident with my answer.

image

This one is interesting because I keep mixing it up with direct (my memory is so awful…wish could add more srs levels and spacing to WK)… but anyway…the vocab again saved me with regards to meaning seeing the 2nd and 3rd in the list. Definitely helped in that regard for sure. Since the mix up is only which kanji is it…(know both readings fine)…it was a 50/50 guess at which one I was seeing. hopefully this will help cement the differences between the two in my head.

image

been having a similar problem with this one (wasn’t a problem before) but now with the added density and with farming … (no idea why) but keep intermixing them all…(really believe if the SRS timing was better this wouldn’t keep happening)…anyway the vocab definitely worked and didn’t miss this one.

one more place there was a script/computer error …absolutely not user error…where it let me down… :face_with_hand_over_mouth:

image
image624×658 38.3 KB

it wanted the kun reading and couldn’t remember it … the script failed me…yes all the scripts fault…

seriously though this is the same sort of problem as mentioned above (borrow/lend/rent)… intent is wrong, the meaning is intention but then mix that up with will, desire, ugh can’t ever remember how to remember which one it is when it’s an idea vs an ordinary noun. Of course I know it now but when it comes up again - and even after posting this the SRS interval will likely be too long and it will be like I know I knew this but it’s completely gone out the window again…and another leech lives on…

anyway… hope you get some entertainment out of my suffering and I know this is still an early experiment in leech squashing so if any of that helps with regards to thinking about how effective it is (or isn’t)…just wanted to share…

1 Like
36

I think I made the same mistake on the same kanji for meaning just yesterday. No, wait, I think it was for lend. Wait, is this one lend?

I did find that in some cases, even with the vocabulary words, I got some of my enlightened kanji wrong. The reason? I didn’t recognize any of the vocabulary anymore!

We have a lot of the same leeches. I think I went too fast and got this one wrong (put “direct”) because I didn’t look at the vocabulary. (“Oh, I know this one!” Well, no, I don’t. That’s kind of why it’s a leech.)

1 Like
37

all signs point to bad SRS intervals… huge problem and effectively makes WK painful…

Had this discussion with several people…Based on several other folks reviews (completely unscientific info) I honestly think most people have the same leeches.

off topic

Given that we can figure out our leeches/values/etc…you would think that WK could actually aggregate the user data and look at the top X number failed/leeches and say yeah these are the most common problems on the platform…a top 100 list if you will

Even a top 10 list would be interesting or a top 10 by groups…so you end up with the top 60… and then when doing lessons/reviews include something that says hey wk peoples pay attention this is one of the hard ones and here’s why we think so…watch out for this and/or that…pay close attention to the radical on the left or in the lower corner…and notice this vs that one… that’s the difference…or this one doesn’t have a mnemonic (it just says it’s the same as blah so you know it…yeah and if you don’t know the one it’s the same as cuz it’s a leech…)…well go back and clean those up… maybe an asterisk or different background color or what not…make the top 10 or top 100 items that are learned and update the list at fixed intervals (i.e… every 6 months)

Others have asked for similar functionality and don’t see WK ever actually doing anything to address the inherent deficit in their SRS system (not addressing leeches globally)…no suspend/ghost reviews
/no additional reviews/fixed single SRS option only … we all just rely on scripts and the handful users who provide free programming to the users to improve the usefulness of WK overall. WK really should be paying you folks!

I’m certain I’d feel different if WK was free (well it’s free so what can you ask for kind of thing)…But we paid for this experience and really wanna learn right? I don’t want to review some of these 263 leeches that I’ve actually already reviewed more than 100x based on the stats…seriously no joke 263 of my 341 leeches I’ve reviewed 100x and failed for various reasons…(I know I’m dumb but it’s super demotivating)…If I didn’t have a lifetime sub I would have quit a long time ago and I suspect that’s one of the reasons many people give up…Languages are difficult and Japanese is super hard, but WK could seriously make it less awful by providing at a minimum some kind of built in leech manager…even just a suspend would be better than doing nothing.

(I’ve said on the boards in the past I’d be willing to be a guinea pig for trying stuff out… but to this day … WK has stayed silent)… sadness…

…walks off soap box…
(really needed to vent…drowning again in leeches)

3 Likes
38

does this work with this other script Simple Show Context Sentence?

I mean since it appears below the kanji or vocab, if it organizes one below the other.

39

I’m absolutely certain WK has this data. I’m also pretty certain they’ll never publicly share it. However the idea to use that data to influence when items show up and how they are described is a great idea.

40

Author of the context sentence script here. That’s a great question. I think it should since at present that are no example sentences for kanji items. But if there were sentences to start appearing, which the context sentence script should be able to handle, it creates it’s own div element and places it above the question-type element. If both were running I presume stack order would depend upon who ran first but it would otherwise work. But if anyone installs both and there’s a conflict, please ping me.

https://the-stack-overflow-podcast.simplecast.com/episodes/one-in-four-visitors-to-stack-overflow-copies-code/transcript/

StackOverflow implemented an April Fool’s prank and in the process discovered that 25% of the visitors copy code.

2 Likes