Reviews complete in only 23 minutes!
And about 30-something sit-ups. I tracked errors per 12 cards, and made myself do an equal number of sit-ups after each session. I think I may have gotten about four right that I otherwise would have gotten wrong, as I was taking a little more time to reduce the number of sit-ups…
Still 34 lessons pending, though. I may come out of lesson hiatus soon. I’m down to 63 Apprentice (of which 59 are leeches…)
Hi @ChristopherFritz, I hope you are doing well 
I’ve been a long time lurker here on your Study Log, which has always been a great source of knowledge and inspiration for me, and recently I was so inspired by your word clouds that I wanted to give it a try myself, so I thought about sharing it here!
Here is my take on a word cloud using all the kanji from the story dialog from Pokemon Black for the Nintendo 3DS.
Following your steps, I decided to take my productive procrastination a little further and also drew this diagram breaking down the kanji by WaniKani level.
So now I know that at about level 35 I should know roughly 80% of the kanji for playing this game–which is important since it doesn’t have Furigana support–which is a goal of mine since I started learning Japanese.
Another thing that I also found pretty cool is that, from the 1257 kanji on the game’s story dialog, WaniKani doesn’t cover only three of them: 嚇、箇、and 謁.
Anyways, I hope you don’t mind me hijacking your Study Log. I guess I just wanted you to know that you are a big reason that I stay motivated to learn every day, and I’m sure I talk for a lot of us here in this forum.
Cheers! 
I like the use of bars here. I’m conditioned to have only one Y axis (numbers on the left, as opposed to having a separate set on the right), so I never thought to combine bars with the line. You get a good feel of how much impact the lower levels have seeing the bars!
Kana-only mode =D
I always find it interesting to see which kanji aren’t covered by WaniKani. For Alpha Sapphire dialogue, it’s nine: 絆, 灼, 蘇, 臆, 楼, 煌, 蓋, 賜, 逐.
I always use unique kanji counts for my charts, but do you have frequency data for the game’s kanji? If so, you can do a chart that will better inform you on your readiness for the game.
Consider Pokémon Alpha Sapphire’s dialogue:
For PokéDex entries across all the games (that support kanji) through Sword/Shield):
If someone waits until they reach a certain WaniKani level before they attempt material without furigana, total kanji will let them know the true level they should dive in. Especially if they’re a slow kanji/vocabulary learner like I am!
I’ll never complain over charts and graphs and number analysis =D
That’s a great point! I do have the frequency data for the game’s kanji, so I was able to make a graph very much like yours:
And curiously enough the results are very similar to your Alpha Sapphire ones: 90% of the total kanji recognition at level 30, but for unique kanji it goes up to around level 45. Pretty neat!
I downloaded a Japanese text dump for the game from the Bulbapedia website, and then extracted only the Kanji characters for all the story files.
To extract the Kanji, I went with this simple regex that (hopefully) matches the Unicode ranges they are in:
kanji_regex = r'[㐀-䶵一-鿋豈-頻]'
I have absolutely no idea as to how reliable this is, since I extracted all of this from random internet resources. But hey, that was a lot of fun to do so I’m happy with it!
I’ve been meaning to check out their text dumps. I got the Alpha Sapphire data the old fashioned way:
(Maybe there were a few more steps involved between 2 and 3.)
I find that a lot of material has a very similar chart. WaniKani staff did a pretty good job of arranging kanji into levels!
I’ve been thinking about doing something like this but in real time, while the emulator is running the game, so that maybe it could run an extension that will dump the text as it appears on the screen (so it’s easy to look up and copy it), or maybe even add furigana to it!
However my experience with roms and emulators is pretty minimal from a coding point of view, so I think it will be easier to just power through until level 30 and play the game the old fashioned way 
but it would be awesome nonetheless!!
It would be neat if a project arranged all the dialogue into where you encounter it in the game, such as “here’s all the dialogue from this town, and on this other page is all the dialogue from this route”.
Then, one could use a tool like the Migaku Browser Extension (still in beta, no free release yet) to see which dialogue is 1T (in this case meaning there is only one unknown word) to make an Anki card from.
But there’d likely not be enough interested people helping out to make such a project work.
Funny enough I was able to find something like this by searching <Name of the game> + セリフ. For example, here is one resource for Pokemon ORAS. Not sure yet how I would go about using this in a practical way while playing through the game though…
I assume this is from your personal Breath of Fire play through? Pretty neat!
I’ve seen you mention this a couple of times recently, and from my first look at their website it looks like it has a lot of potential to be a productivity boost on language learning. Still haven’t tried it personally because I feel I’m juggling too much stuff at the same time at the moment, but I’ll get there!
If only it had the kanji!
With kanji + arranged by location + Migaku Browser Extension, you can easily see your known and unknown words.
For example, taking a random bit of text from Alpha Sapphire, I see:
The red underlines signify words I don’t know. (Or haven’t marked as known yet, such as エナジー.)
Looking at this, I would see that there are too many unknown words for me. I wouldn’t recognize enough of the kanji to even be able to read it, and I don’t even see a single 1T sentence. (If there was a sentence with only one unknown word, the word would be highlighted in yellow.)
(You don’t really need it to be arranged by game location for this, but such arrangement makes it easier to pre-learn words before playing a section of the game.)
Yup.
(Ah, I just noticed I’m missing a line from the last row. Gotta go fix that…)
The Migaku Browser Extension is something you want to be using if:
Right now, my main immersion is manga, so I’m not using the tool to its fullest yet.
Today’s distraction: Chatting with rafascar Writing software!
A lot of my work time is spent writing software. But I’ve had a project lately that has taken up a good portion of my time, meaning I’m writing less software at work for the moment. I think this has left me with a need to write software on my own time.
My programming project for this Saturday has been to refine my word frequency stats. For example, if I’m reading a book, and a word I don’t know comes up, should I make an Anki card for it?
This involves a few things:
The summary output isn’t pretty, but it’s usable:
I think there’s a plan for the Migaku Browser Extension to eventually allow loading in user-selected frequency lists. The ones my code is able to generate should be easily adaptable to that. Then when I go to make Anki cards from a source, such as from an episode of the anime Saint Tail, I’d be able to see the frequency of that word within the whole series, as well as among everything I’ve run through my program.
Today’s distraction: Migaku’s kanji add-on for Anki.
Their kanji add-on just went from alpha release to beta release, which means it’s gone from Patreon US$10 tier availability to Patreon $5 tier availability. (Migaku makes its programs/add-ons free once they are out of beta.)
I knew in advanced (based on the add-on’s Discord chat) that it would be centered around mnemonics. I think it’s been determined many times over by now that mnemonics are about the best way to learn kanji for people learning Japanese as a second language. (Too bad for me that mnemonics pretty much never work for me…)
I don’t plan to quit WaniKani any time soon, but this add-on does add in some things I’ve wished I could do with WaniKani:
1) Learning kanji in my own determined order.
One of my mined sentences is for the word 将来. WaniKani teaches this word at level 36.
Currently, I’m one month into level 29, with the following progress:
I’m constantly encountering words in manga reading that I’ve “recently” learned in WaniKani. Big plus. No complaints there.
But there are also words from WaniKani that never come up in what I’m reading. I have burned vocabulary words that by the time I do see them in reading, I don’t remember them even after looking them up in WaniKani.
Migaku’s kanji add-on lets me select my 将来 card, and it generates radical and kanji cards for the unknown kanji on it.
This allows me to tailor my kanji-learning to the words I’m encountering when reading (including watching Japanese-subtitled anime).
2) Suspending leeches.
One reason I’ve spent the last month with almost zero progress on level 29 (after spending 30+ days on each of the past few levels) is because I spend so much time with the same leeches over and over. Even now, my Apprentice is 75 leeches out of 77 cards. (My plan to auto-pass any Apprentice card didn’t pan out, as I keep forgetting to.)
Right now, over 90% of the time I spend reviewing cards in WaniKani, it’s the exact same cards I’ve been reviewing every week for months and month and months. (And months.) Some cards I’m probably seeing every day, and I still don’t remember them.
Since Migaku’s kanji tool is an Anki add-on, any kanji card that becomes a leech can be set to auto-suspend. No need to spend time on it anymore. That time can be better spent learning other kanji, and I can return to the leeches when I start learning other words that use them.
A negative of Migaku’s kanji tool, for me, is that it’s based on Heisig’s radicals, with no option to use WaniKani’s. I can manually add in WaniKani’s (takes time and effort), which allows for consistency, but then laziness will surely set in. (Not that mnemonics have ever really served me well to begin with…)
Edit: After a bit more than half an hour of watching GochiUsa season 1 episode 1, I’m about six minutes in with six mined vocabulary cards, and 52 new radical/kanji cards that I’ve gone through. Now to see how reviews for these go over the next week before I start mining any more!
These cards from the Migaku kanji tool include kanji I know, but I’m letting it add cards from scratch so I can get an idea of the default radical names. Depending on how things go, I may come up with my own names for some, or add WaniKani’s radical names for some. Not that I expect mnemonics will magically work for me this time, but no harm in giving it another go.
A post it note on your monitor?
You could sort of replicate this on WK by just passing the leech, no? Then there’s no demand on recall and that’s basically setting the card aside. Even if you artificially pass something, since you’re now using another system, the kanji will show up eventually and the vocab, maybe, so I don’t think you’d be losing anything.
Or, are they from a particular level, and resetting a few levels might be useful?
It might be to your advantage to play around with the SRS a little. Oh, and the BishBashBosh site is super helpful! I am finding the extra run through super helpful.
Unless there’s some trick that I don’t know 
(Post-it note won’t work when I do reviews on a smartphone at different locations depending on the time of day. And if I were reviewing on my desktop computer with a post-it on the monitor, it would likely become invisible to me during reviews.)
I keep forgetting to use it. I used to rely on @Rowena mentioning it in their study log as a reminder.
You’ve got me curious now. I haven’t checked leech stats in a while, and I never looked into levels.
Excluding my smaller leeches, here are the ones that show up the most often in my daily reviews:
If I reset to level 8, I’m golden =D
Actually, there is hope for me yet!
I’ve been writing some tools to track words in subtitle, book, and game texts that I can use for reading. I can now easily look at the frequency of my leech kanji/vocabulary in materials I have available, and find scenes/sentences that use them! Granted, they may not be 1T sentences, but a 2T or 3T sentence (meaning two or three unknown words) can easily become a 1T sentence if the other unknown words are easy to learn.
My biggest vocabulary leech is my level 9 仮定, currently sitting at Apprentice I. I’ll do a search of my word lists to see where it gets used, and how much:
0 matches
Okay, that’s a bad example.
Let’s see about my second-biggest vocabulary leech: 狭める. This one I know the meaning (as I’m familiar with 狭い), but I almost always fail on the reading.
Sometimes I think the reading of 狭める is せばめる. You’d think that means I’ll get it right. Except I pause. I think, “I always get this one wrong. If I’m thinking it’s せばめる, then I know I am wrong, and I can purge せばめる from my memory.” Thus I go for a different reading, and…that’s why it’s Apprentice II right now.
(I can only tell you that this is せばめる because I reviewed it this morning. And I think I reviewed it yesterday as well. And a day or two before that. I’ve seen it a lot lately. But I’ll forget it sometime before the Guru 1 review.)
My list of over 15,000 words extracted from the aforementioned sources shows that 狭める comes up this many times:
0 matches
Well, that’s a word I won’t be getting a lot of mileage out of…
I’ll try searching for just the kanji and see how many results I get. If I can find a source I’m familiar with that has a different word with the same reading, and its in a 1T sentence, I can make a card from it.
Note: This excludes 狭い because I have it on my known list.
It looks like one of my biggest leeches also doesn’t come up in anything I’d be reading (aside from maybe manga, which isn’t included in my lists).
Maybe I’ll have better luck with…
Okay, I know this one. It’s ひりょう. Or しりょう. (ichi.moe tells me it’s しりょう.)
This one actually has hits on my frequency list!
名詞 資料 32
Checking where it appears:
I haven’t checked to see if the subtitle files I have for Cardcaptor Sakura will work with my DVD’s.
But the second highest frequency is the Detective Conan book I’m currently reading! I haven’t reached the first one yet:
It looks like it’ll be paired with 捜査, which I feel like I know on my own, and haven’t seen in WaniKani yet. (It’s a level 25 vocabulary… Coming up for burn review in 11 days!)
Looks like I’ll have two candidates for sentence mining (items shown with a yellow highlight):
Okay, back to manga reading, then on to tonight’s Conan reading. Not that I’m anywhere near those ひ-or-しりょう’s yet.
Edit: 資産家 comes up three times in episode one of Saint Tail. I may just make a card for that for extra 資 reading practice!
Today’s distraction: Nothing! Just a plain, ordinary log update.
Reading the Detective Conan 「瞳の中の暗殺者」 movie novelization is going well. I haven’t kept up my goal of reading 2% every day, but I’ve kept up a 2% daily average. (Example: I missed reading yesterday, so I read 4% today.)
So far, this has been the easiest book I’ve tried to read. Today’s 4% was especially easy somehow. I don’t get all the words. Sometimes I look them up. Other times I don’t. Just depends on whether I feel I have a basic idea of what’s going on or being said.
That reminds me, 張り込み came up five times, and I persisted in not looking it up. I felt pretty confident in its meaning from context. Perhaps I’ll check it right now. Let’s see… Ah, yup, “stakeout”. That’s what I figured.
I’m up to 37%. Past the one quarter mark, but not at the halfway point for another week yet.
I’ve slowed down adding new kanji via Migaku’s kanji add-on, as I’m currently sitting at 108 cards created across four days. Cards counts added per day have been:
| Day | Cards |
|---|---|
| 6/13 | 52 |
| 6/14 | 8 |
| 6/15 | 25 |
| 6/16 | 23 |
| 6/17 | 0 |
Some of the radical and kanji cards have been easy. Others are the same as I experience with my WaniKani leeches, where even though I reviewed the card just yesterday, today I could swear I’d never seen it before in my life.
One thing I didn’t try with WaniKani was coming up with my own meanings for radicals. There were two reasons for that: 1) I might assign a meaning to a radical that I can’t work with for various kanji mnemonics later, and 2) I lose the ability to use WaniKani’s mnemonics for every kanji based on that radical. I suppose there’s a bit of 3) the radical would lack any relation to the kanji’s it’s used in (but that’s true for WaniKani’s radicals, and the Heisig ones Migaku’s add-on uses).
Since mnemonics never really worked for me (whether via Heisig’s Remember the Kanji, or the Kodansha Kanji Learner’s Course, or WaniKani, I’d say item 1 has been the least of my issues. For item 2, since I expect a lot of the kanji I’ll be learning from the add-on isn’t covered by WaniKani, that’s not too big of an issue.
I’m considering going the “make my own radical name and mnemonic” route, but I’ll wait a little more and see how things go.
I’ve mostly suspended mining new sentence cards this week. This suspension is planned to last until I see how things go with the Migaku kanji add-on.
Right now for my sentence card, I’m concerned only with recalling the meaning of the word, regardless of whether the kanji or its reading triggers that recall. One effect of this is that I don’t learn the kanji. I’d like to transition this to having kanji cards created when adding a sentence card, and then reviewing the kanji cards before reviewing the new sentence card.
I had downloaded a few subtitle files for となりのトトロ, but none of them were quite right. Maybe there’s a good one out there, but the ones I found had multiple lines of dialogue shown at the same time (one above the other), rather than timing them separate, and also included names for off-screen speakers in parentheses.
I ended up taking an English subtitle file, and editing it to substitute in the Japanese dialogue.
I thought this would be relatively quick and easy, but it turns out that even then I needed to split many lines for a better result. And I couldn’t get any way to easily re-time the split lines, so I was just manually adjusting numbers, refreshing the subtitle file in MPV, and re-playing the line until I felt it was about right. Sometimes an English line was a little longer or shorter than the Japanese, requiring adjustment.
And because I like to make things more difficult than necessary, I just had to set different font colors for each speaker.
I actually started doing this over the weekend, but only finished up today. (Unless I decide to kanjify some more words, as I’m using this file with Migaku for sentence mining.)
Although it was a massive time sink, I’m quite happy with the result, and I more or less enjoyed the process. It also gave me the opportunity to catch a lot of little things I’d never noticed before, like brief facial expressions, and audio I’d never noticed (such as the sound of the buckwheat hulls when Satsuki plops down onto her pillow). So maybe it was an okay use of my time.
This weekend, about 80% of my Anki kanji cards went from previously being “I know this every time I see it!” to “I’ve never seen this before in my life.”
It quickly became a huge time sink, so I went ahead and wiped out the whole deck. I need to rethink how I go about adding cards to it.
Using my recently-written tools for analyzing word frequency for anime I’ve watched and some Pokémon video games, I’ve extracted kanji frequency. Then I went down the list of kanji, and looked at which ones I know.
Some of the kanji I knew instantly. Others I thought I knew instantly, but it turns out I confused them with a similar-looking kanji. Others I thought I knew instantly, but I was completely off. And this is among the 100 most common kanji one is likely to see. It’s not uncommon for me to question the meaning of a kanji I’ve encountered weekly for the past three years.
Clearly I need to really return to the basics.
I’ve done eight lessons in the past 26 days, and my last was one lesson fifteen days ago. I’ve instead focused about an hour a day on WaniKani reviews, doing about 50 to 150 reviews per day.
Following my early morning’s 72 reviews, my Apprentice made its way back up to 100. Out of my 66 reviews for late morning, I’m back down to 80 Apprentice. Following my afternoon and night reviews (another 40 or so), I’ve settled on 86 Apprentice for the night.
This seems to be the common range for the past month or two, during which time I’ve been trying to get the average down to 50.
During my reviews today, at least a third of them felt like “I’ve never seen this before in my life” and another third were “I know for a fact this is x” only to find it’s not.
I’m sitting at 475 leeches. This is spread across:
Over half of the leeches are for levels 20-28.
If I dropped back to level 20, my leeches would drop by 54%, down to 220 (still a large number). This would leave the spread as:
That would bring my overall Apprentice down to around the 30 range, and my daily reviews would likely fall from the 100–180 card range to the the 20–50 cards range.
The idea of dropping my level and reclaiming my time, to spend on learning kanji and vocabulary (versus reviewing kanji and vocabulary) has become increasingly appealing this past month.
The spread of the remaining leeches would be:
I don’t know how I would go about tackling these, even at a reduced level.
The most stand-out thing for me to discover recently is that a lot of my vocabulary leeches do not appear in material I consume or am likely to consume.
Out of the material I’ve done text analysis on, looking only at verbs, adjectives, nouns, and so-called na-adjectives, I have about 450,000 words analyzed.
The most common word is する, showing up 15,375 times. 知る shows up 997 times. 新しい comes up 301 times. 光る comes up 103 times. I think anything that comes up at least 100 times in this data, I have a fair chance of encountering.
The following chart tells how many of my vocabulary leeches show up how many times:
(Left side: Shorter bars are better. Right side: Taller bars are better.)
That first blue bar on the left says that of these 293 vocabulary leeches, 101 do not show up at all. That’s 34.4%.
Another 111 vocabulary only show up 1 to 10 times. That’s another 37.9%.
And 66 vocabulary show up only 11 to 50 times, for another 22.5%.
That means of these vocabulary leeches, I’m not getting native material exposure to more than 95% of them. That’s…a lot.
I should mention, this is only my leeches. I’m constantly seeing words in manga that I recently learned in WaniKani. And most of those vocabulary don’t become leeches.
This is the number one reason why I wish I could suspend leeches on WaniKani. That would free me up to continue learning new words, without being bogged down by the same leech reviews every day. Then after getting to and through level 60, I could revisit all the leeches for “round 2”. (Or not, if I wanted to focus on my vocabulary frequency list’s kanji not covered by WaniKani.)
For the longest time, I’ve thought it would be nice if WaniKani would let me choose kanji/vocabulary I want to learn (based on what’s available on the site), and it would queue up those and their prerequisite items.
I can get something similar from Migaku’s new kanji add-on, although it’s kanji-only. I’ve started using the web site Tatoeba for sentences to go with the kanji, and I’m focusing first on the most common kanji that I still have issues with.
Once I’ve gotten the basics sorted (hopefully to the point I don’t second-guess myself on recognizing kanji I encounter in reading every day), I’ll assess moving forward with new kanji. And when I do, I’ll have my frequency lists to help me find the high volume ones I’m personally likely to encounter to begin with.
If I haven’t dropped my WaniKani level by the end of the week, please send help.
Amen!!! Leeches are hell… WK really needs to do something built in instead of relying on external scripts and everything else…(how many users have been complaining about this for years… )
honestly what I’d love is a dynamic SRS and if not that, more review cycles (if they are going to go around for a year anyway)…might as well take a year with proper intervals instead of making leeches …
What has helped me currently is that don’t always do new lessons everyday. I only do new lessons based on the quantity of reviews the following day…and an assumed my fail rate (don’t know if this will help or not but you can certainly see about something like this)…my average % correct is around 73-74…so I assume a 30% fail rate and add that number to the next days reviews and decide if that’s going to push me over 100 the next day… it isn’t perfect and sometimes I’ll go slightly over but it does keep the reviews manageable (with the leeches that are). Then I’ll add 3-8 vocab items (sometimes 10 but almost never more than that)…and with new lessons watching carefully as doing 10 at a time will quickly prevent forward progress (for me). I typically check in the AM and if there are 70+ reviews, I wait until about 10am or so and then re estimate before adding lessons.
In the past, I have found that every 4-5 levels I have had to stop and do no new lessons for 3 months or so and just grind out reviews until the apprentice pile falls below 30 … then I’d start doing lessons again. I’m hell bent on NOT resetting (feels like giving up)…or I might have reset by now… I’m kind of right at that break point again though level wise for leech churn… would have already leveled up but doing new lessons with the new junk wk is adding is slowing me down more than I already am going…
From this…not sure but trying to see about managing reviews…so far just started (after 28 days) level 32 radicals… it’s the new junk wk is adding … but going to see if this method works for a bit… maybe I won’t have to stop doing any new lessons, even if the monthly level up cycle gets pushed out … my current numbers are
if it gives you some basis of comparison…
Agreed…exposure really does help!!! Especially when the word pops up in something immediately and repeatedly in native materials. And you read a lot! If only WK would do something about leech management …(it’s an inherent problem in the platform that really needs to be addressed at a global level)… we all don’t have perfect memory…I’d take WK addressing the leech problem over adding new vocab and changing the interface around any day…
One thing that would be a really interesting statistic (don’t know if we could figure it out)…but what are the top 50 (or maybe 100) leeches for ALL users and what is their frequency in the wild (for average folks). I’m sure WK has that info in the DB] In any case you’re good at doing those sorts of fancy number things and making them pretty…while it would be nice to learn everything…maybe there’s a way to figure out what can be truly ignored while slowing through WK.
Don’t feel too badly about the leeches…I know some level 60 folks and my leeches turn out to be the same ones they have too… heaven help me if they stick around for 2-3 more years though… might throw WK out the window if that happens * ┻━┻ ︵ヽ(`Д´)ﾉ︵ ┻━┻
Always interesting following your studies…good luck!!!
I’m all about those long levels!
(Note: I originally signed up for WK in 2015, used it a little, then stopped maybe at level three or four. When I returned for 2019, I reset to level 1, thus the reset shown above.)
My current dilemma is I haven’t found anything that works for me for doing this. I’m always trying out different things, though, and I may one day stumble upon something that works for me!
My review fail rate ranges from about 30% to 80% =(
But I haven’t looked at how many vocabulary have become leeches versus not.
I’ve trying for some time now to get my daily reviews below 100. I’m currently around the 80 to 160 reviews per day range, judging by recent days. Although being two reviews each, reading and meaning, it’s more like a goal of 200 reviews per day, where I’m doing 160 to 320 per day.
My non-WK reviews today will be 56 reviews, but I expect that to go up as I add more kanji and vocabulary cards to refresh myself on basic kanji. (Sentence mining from consumption of native material is still on hold.)
I’ve gotta admit, that apprentice number is enticing, and makes me want to drop my level just so I can get something like it =D
Edit: Forgot to mention:
For me personally, I view resetting as a strategy. There are different reasons someone could reset. Whether it’s a good strategy or a bad strategy depends in part on the reason (high number of reviews; high number of leeches; poor retention on recent levels, etc).
I wouldn’t say it works, but just not doing any lessons for 3+ months…eventually most of the enlightened items all work their way back around and eventually it falls on it’s own…it’s discouraging but eventually I get there… (have you ever tried not doing any new lessons that long?) it does make a big difference when you stop adding new.
The reason I suggest this is not to say don’t reset, but if you haven’t tried it…you might not need to actually reset if you can work off most of them over 3 months…will depend on how far back resetting would put you vs spending the time grinding … don’t know what’s less painful though…really you have to decide…(it takes serious will power not to add any new lessons at all - it’s frustrating feeling like you aren’t moving forward - feels like everyone is passing you up)
every once in a while I get a bad set…but overall it’s right around low 70s… this morning 70% exactly with 44 reviews…
I found that eventually even the awful leeches sometimes would churn at app level 1 for sometimes two days…and eventually they would move up…(super infuriating though…missing app1 over and over)
Never meant resetting was a good or bad strategy…for me personally I think it would be bad…(I’d feel like I was giving up) after all even if the card is a fail…it’s not like I haven’t seen it before (even though it often feels like it’s the first time ever seeing it). If resetting will help…then do it…don’t hesitate…kill the leeches…if it won’t help then don’t…[as long as WK doesn’t have leech management…could always abandon WK altogether… use anki and then you’d have the control you want, suspend leeches, add words/kanji in whatever order you want, etc…] … if I didn’t have a lifetime sub…I would probably have done that along time ago… and I h8 anki… it’s so boring hehe
Looking forward to seeing how things work out… Best of luck on whatever you end up doing… 
I’ve gone long bouts without lessons, but definitely never for three months.
I actually 100% agree with your method as a route for me to take…if I hadn’t started seriously using Anki recently.
With Anki on my plate, I’m asking myself the question: how can I best learn new words, without being bogged down by leeches?
As things are with WaniKani, I’m unable to do that.
I don’t plan to abandon WaniKani completely for SRS. (At least not yet!) But I may just transition to learning new words completely on Anki, and let WaniKani be a leech review platform for a while. (Possibly on a lower level to free up time for learning new words on Anki.)
For that front, I have two paths in front of me, and ultimately I may use a mix of both:
Create cards for the highest frequency vocabulary on my personal frequency lists. That way when I encounter them in reading, I already have an awareness of them, and can more easily learn them.
Create cards for words I encounter while consuming native material.
This is what I was doing for a while, but I need to improve it so I’m only adding higher frequency words. I don’t want to take up my Anki time reviewing words that I may never see again. (For this, I need to parse more subtitle files for anime I’ve seen that covers the same genre/domain as the manga I read, to get more accurate frequency stats. Simple enough to do.)
For both of these, the goal for this is to learn the reading+meaning, but not necessarily the kanji. I’ll rely on WaniKani and/or the Mikagu kanji add-on for Anki for the kanji part. I’m hoping that learning words in advance of the kanji they use will make the kanji a bit easier to learn.
Edit: Also, I may mine 1T sentences from Tatoeba for my WaniKani leech vocabulary, and make Anki cards for them. Maybe try that before dropping any levels.
