It has seemed as though you have been spinning your wheels with leeches for awhile so it’s cool that you are problem solving for success. Plus, with all of the tools at your disposal it kinda seems like WK isn’t the best learning tool (for your needs) at the moment.
Would you ever consider just entering in the correct answer for those leeches with the caveat that you would fail or resurrect them months down the line?
I would keep forgetting to.
That was half of my planned strategy: auto-pass all Apprentice items to reduce my daily reviews, so I can take on new lessons (that hopefully don’t become leeches!)
If I had a way to auto-pass the leeches, I’d need to track them to resurrect later. Not an insurmountable issue but it does add some friction.
I have wondered about writing a script that I can run daily that would query my due cards, and mark anything on a “list of leeches” text file as “correct”. That would solve the issue completely: I’d have a list of everything being “cheated” through, so I could resurrect them later, and I could focus on new words.
And now I’m pondering such a script that could simply do that only for Apprentice and maybe Guru reviews. Auto-pass them, but let them still come up (and likely be failed) on post-Guru reviews, just so I have them still in mind. And if I do Tatoeba sentence cards for them in Anki, maybe that’ll help me learn them so I don’t fail post-Guru reviews in WaniKani. And since nothing would auto-burn, there’d be no need to resurrect anything.
Hm…
Well. Here I go.
Visual representation of my code marking Apprentice vocabulary leeches as “correct” without me reviewing them:
And the result:
I’ll just need to add in kanji, and that’ll wipe out most of my Apprentice, freeing me up to learn new words. But for now I’ll still be hit with the leeches at Guru I review time.
Edit: After manually doing the rest of the reviews, my Apprentice…went back up by five.
even if it doesn’t work it sure is hella cute!!!
This is what I imagined you’d do when I mentioned the “caveat” 'cause you’re so on top of data…
(After experiencing the fluffy cat cloud upthread, I made my way over to @ekg’s thread and helped myself to some cat stuff. )
Oh, I love this idea! 
It makes leech squashing even more fun for sure! 
I hope it works! 
I updated my list of auto-pass vocabulary leeches to exclude higher frequency words (as per my frequency lists).
Even if I’m constantly making mistakes on vocabulary such as 現れる, 捜す, 自由, and 難しい, these are still fairly common words.
Especially 現れる. I kept seeing it when reading through Sailormoon, and I kept having to look it up. I think I was looking it up basically every week for several months before it started to stick. Yet when it came up in my WaniKani vocabulary, it quickly become a leech. I feel like I’ve known it fairly well for several months now, and yet somehow it’s only Guru I. How?!
I wonder if when I keep reviewing the same leeches over and over, if I get into a rhythm. Like, if I see 自由 and mistakenly think りゆう, maybe there’s some pattern matching that has formed that makes me think 現れる is みだれる or something.
I actually honestly have no idea what I mistake 現れる for that leads to failed reviews, even though I’ve been reviewing it probably nearly weekly for the past 10 months (not to mention the months of exposure while reading manga before that). And that not knowing sounds like it’s a symptom of the unidentified problem.
If it’s an actual thing where I’ve pattern-matched the same mistakes together, I wonder if removing a bunch of leeches and adding new lessons into my reviews can break that, allowing me to start getting some of these year-long leeches right. This is just speculation, though, but I’m writing it out to help keep it in mind to watch for.
I’m not planning on doing lessons just yet. (Except for the low-level ones that were recently added.) All these auto-passed Apprentice leeches will still come back for Guru reviews. Also, I want to put more time into reviewing my highest frequency kanji in vocabulary cards in Anki, so there’s where I want new lessons to show up for now.
As those Apprentice shrink, so grows the Guru. I’m both looking forward and not looking forward to when the leeches come back at me as Guru.
Over on the Anki side, it’s difficult to track how many common kanji I’ve added this week, since adding kanji also auto-adds cards for the kanji’s radicals (via Migaku’s kanji add-on). Kanji + radicals together make up 28 cards in six days.
For some cards, I’ve settled with the default mnemonic keyword. Others I’ve created my own. Others I didn’t like the default, and couldn’t come up with one of my own, but I know the kanji well enough that I’m trying that out as my keyword. I’ll see how that goes.
After creating the kanji card, I also take my vocabulary frequency list, and add cards for the most common vocabulary using that kanji. That deck is up to 52 cards. So far so good there, although I’m missing the imagery and audio that I get from anime sentence cards. (These kanji vocabulary cards do have audio from Forvo of the word being read.)
My initial goal is to get through all kanji that show up at least 1,000 times across the sources I’ve run a text analysis on. This gives me 105 kanji to focus on initially. It’s a nice starting place because aside from them being the most common kanji I’ll encounter, I’ve already covered many via WaniKani.
I even have 藤 as number 94 on the list. Who knew it was so common? WaniKani gives us 藤, and calls it a day. My frequency list has:
| Word | Frequency | Notes |
|---|---|---|
| 藤宮 | 616 | Name. It’s amazing how often Fujimiya’s name comes up in the one-season anime for One Week Friends. |
| 藤隆 | 175 | Name. I didn’t realize Sakura’s father Fujitaka’s name was used so much in Cardcaptor Sakura. But the (original) anime did have a long run. |
| 佐藤 | 95 | Name. Satou is a character name in a few series, but mostly this one’s from the Detective Conan movie book I’m reading. |
| 後藤 | 90 | Name. Somehow this character from Kokoro Connect has this kanji for the name is read Gotou. Funny thing, this voice actor’s name is Fujiwara, using 藤. |
| 藤岡 | 59 | Name. Ouran High School’s Haruhi’s family name is Fujioka. |
| 藤島 | 37 | Name. Kokoro Connect’s minor character, Fujishima. |
| 工藤 | 18 | Name. Detective Conan’s Kudou, giving the same “tou/dou” pronunciation as Gotou above. Also, a minor one-episode character by the same name in Saint Tail. |
| 伊藤 | 11 | Name. Itou, a surname for a couple of main characters in Strawberry Marshmallow. Also, a minor one-episode character by the same name in Sailormoon. |
| 藤原 | 11 | Name. Fujiwara design office from Honey and Clover. |
| 内藤 | 8 | Name. Naito, a minor character from Aoharu Ride. |
| 遠藤 | 7 | Name. A character named Endou. |
| 加藤 | 2 | Name. Katou from Porco Rosso. |
| 斎藤 | 2 | Name. Saitou, a student in an episode of Card Captor Sakura. |
| 葛藤 | 1 | It’s…an actual word! かっとう, meaning conflict. |
I don’t expect anyone to have read any of that table (sorry if you did…), but I feel I’ve become a little closer to the readings of 藤. Too bad I never learned the kanji beyond “Hey, it’s the most complex-looking kanji I’ve encountered in WaniKani, up for review. It must be ふじ.”
I suppose I can’t say I wasn’t warned.
I confuse this with 現す way too often. 
After looking up it up myself, I wonder why WK didn’t include “Fuji~” as a meaning/use. Because aside from plants species and 葛藤, it’s just in names and names of things.
When a verb ends in す, I think of the すin する. You’re doing something (as opposed to being something, or having something done to you). So for the verb that involves appearing, the “do” version is to “make something appear”. Or, simply put, “to show”.
As far as I know, this logic works for any verb that ends in す. Here are a few examples that do:
| Verb | Action |
|---|---|
| 乱す | to throw (something) into disorder |
| 乾かす | to dry (something) |
| 亡くす | to lose (someone) through death |
| 交わす | to exchange (something) |
| 任す | to entrust (something) |
| 侵す | to invade (somewhere) |
| 促す | to urge (someone) |
| 写す | to photograph (something) |
| 回す | to rotate (something) |
| 慣らす | to train (an animal) |
I don’t know if す-ending verbs are always transitive, but they seem to be doing, rather than being or receiving.
If WK applied that logic to every common name though, a good portion of the first 10 levels would have the Japanese as an acceptable meaning 
I like to garden and admire flowers, but the odds are that I would see the kanji as part of a name before using it in some sort of discussion about botanicals.
But I get what you’re saying… it’ just that for a kanji like this with such limited vocabulary, listing a more common usage (or “meaning” if you will) isn’t a bad thing. Like WK does for vocabulary like 河童 or 怪獣. (These are on my mind because I’m watching a lot of anime recently- jury is still out on the newest Godzilla). But what this all comes down to is the usual debate about how to set aside questioning 
vocabulary choices when the site is mostly focused on the kanji and readings.
The difference with those particular examples though are that 河童 and 怪獣 are specifically Japanese. Wisteria/藤 is native to many places including North America and China. Personally, I was familiar with it in English as a plant long before WK.
Even though some kanji might be common in people’s names, there’s no guarantee that it’ll be pronounced like it is in words. From my observations, Japanese people will write out their name in kanji on important documents, but when needing to write down each others’ names in the absence of business cards, they write them in kana.
Additionally, place names do often relate to the things that are there. Like if I saw 藤 in the name of a place, I would think that there are some wisteria plants there and unless it’s historically divorced from its origin for some reason, it’s pretty reasonable to expect the name to relate to the place.
For example
北九州 is the northern(most?) city in Kyushu, 長崎 is indeed a long cape, and 青森 is fairly forested.
Sorry @ChristopherFritz to keep hijacking your thread.
I’ll just leave this here: my point is that WK includes the romanizations for vocabulary probably because whatever it is is pretty common to non-Japanese. My point about 藤 is that it has a common usage in Japanese (based on Chris’ investigations) in names. Specifically in literature where you’ll find the kanji with furigana, as opposed to the IRL instances that you mentioned. That said, if you ask my husband (who is Japanese), he would say that more and more Japanese, once they’re out of school, don’t know/remember how to write the kanji (in this era of phones) which is why everyone uses kana. But he’s pretty old school. But not old school enough because he definitely needed his dad’s help to fill out some government forms 
… And he’s from 宮崎県 (southern part of Kyushu).
Anyway, obviously WK chose not to close the information gap by only referring to “wisteria” as such, and only including the one reading. Which is fine. I often find myself digging deeper into vocabulary because sometimes the meanings that WK chooses for kanji don’t make sense in vocabulary form. Usually because more relevant meanings haven’t been included. Such is the nature of language learning.
No problem on the thread hijacking.
My text analysis will of course be completely biased to the type of material I’m likely to read/watch, but I do enjoy these frequency numbers:
| Word | Frequency |
|---|---|
| 藤 | 8 |
| ウィステリア | 19 |
=D
According to my dictionary, and I also glanced at Jisho, it’s 291 out of 2500 most used kanji in newspapers! Who would have thought?? Of course now I am super curious, but since I don’t have computer skills to run through data, I’ll just have to read more news to see in what form it shows up.
I think it’s because 佐藤 is the most popular family name by far, 2% of the population has it, so it’s bound to show up a lot in news.
I think WK has a pretty clear streak of only including name specific readings for major companies and places though. I’m losing track of what we’re discussing a bit to be honest. I think I’m just having a stubborn brain fart at this point.
Well, not yet. I feel I’m getting there, but let’s not jump to the end of the story.
Today’s entry is on how I’m going about making Anki cards to help improve my recognition of the kanji I should already know.
I think I’ve mentioned before, but I wanted to do a post putting this all in one place.
There’s a lot of common kanji that falls into one of the following for me:
I’m barely into level 29, but I’m close enough to “halfway through WaniKani” that I feel like this is a good time to get back to the basics and improve and solidify.
I’ve written Ruby tools that extract words and kanji from various text-based materials that I’m either likely to read (such as an anime I’ve seen with English subtitles that I’d like to re-watch to Japanese subtitles), or materials that I’m unlikely to read but are in a genre I am likely to (such as anime I’ve seen with English subtitles, but I probably won’t re-watch with Japanese subtitles).
The kanji I’ve put into a spreadsheet, sorted by frequency.
Here, if there’s a kanji I know the meaning, on reading, and kun reading, I can make its known status as “true”. If I know at a glance it’s a kanji I have trouble with, I mark the known status as “false”. If I’ve created Anki vocabulary cards specifically for learning that kanji better, I mark it as “Anki”.
I also track if the kanji is one of my leeches on WaniKani.
The total column tells how many times the kanji has come up in my analyzed sources. The percent column is completely unnecessary, and I’ll probably remove it. It just tells what percent of the overall kanji appearances that kanji accounts for.
From there, I’ve focused on the 200 most common kanji, out of a complete list of 2,915 kanji. That’s 6.9% of the overall unique kanji on the list that I’m focusing on.
Each day, I look through the list and try to find one or two kanji that I’d like to improve my recognition of.
Today, I’ve selected 考 as my kanji. I know it’s used in 考える, but offhand I couldn’t tell you its on reading.
Apparently I knew it once upon a time:
I don’t recognize 考古学 one bit, having burned it one year and eight months ago.
When choosing vocabulary words, I disregard whether WaniKani includes the word or not.
I pull up my list of words from my text analysis results, and pull a list of all words that include the kanji. For 考, this is:
| Part of Speech | Word | Frequency |
|---|---|---|
| 動詞 | 考える | 1503 |
| 名詞 | 考え | 289 |
| 名詞 | 参考 | 49 |
| 名詞 | 考慮 | 17 |
| 名詞 | 思考 | 13 |
| 名詞 | 考古 | 11 |
| 名詞 | 熟考 | 8 |
| 名詞 | 選考 | 8 |
| 名詞 | 考察 | 8 |
| 名詞 | 黙考 | 5 |
| 名詞 | 再考 | 3 |
| 名詞 | 考査 | 3 |
| 名詞 | 考証 | 2 |
| 名詞 | 考えちがい | 2 |
| 名詞 | 考案 | 2 |
| 名詞 | 一考 | 1 |
| 名詞 | 考え違い | 1 |
| 名詞 | 考 | 1 |
I drop off everything below frequency 10 first thing. If the list is still fairly long, I’ll trim off some more. I don’t necessarily aim to make the list real short, as I may not use all the words near the top of the list.
For 考, I do happen to end up with a fairly short list. Within that sub list, only the first three results are appealing to me (based on the frequency number), but I’ll also check the others.
| Part of Speech | Word | Frequency |
|---|---|---|
| 動詞 | 考える | 1503 |
| 名詞 | 考え | 289 |
| 名詞 | 参考 | 49 |
| 名詞 | 考慮 | 17 |
| 名詞 | 思考 | 13 |
| 名詞 | 考古 | 11 |
My first step is to create a card in Anki. For this, I use Migaku’s dictionary to a word translation, maybe an image, and especially an audio recording of the word being spoken.
Next, I head over to Tatoeba for a sample sentence. I try to find one that has only words I know, but if that’s not possible, then I’ll grab the English translation as well.
If there are no sentences at all for the word, then I’ll do an anime subtitles search for something suitable that can stand alone out of context.
I’m in luck today, because I found the following sentence:
「最近考えることが多過ぎる」
The day before yesterday, 最 was one of my added kanji, and 最近 one of my vocabulary. So having it appear in a sentence on another card is a bonus.
I actually don’t need a card for 考える, but I want be 100% certain of that, which will happen by my getting it correct on many reviews. I did opt to skip making a card for 考え.
As for the other words:
| Word | Notes |
|---|---|
| 参考 | As I figured, Sherlock Holmes accounts for 31 of the 49 occurrences of this word. If I didn’t vaguely somewhat recognize 参 (burned it one year and three months ago), I might have skipped this word as being low frequency outside of one series I don’t plan to read in Japanese. |
| 考慮 | This one is 14 out of 17 appearing in Sherlock Holmes, and I’ve never seen 慮 before. I’ll pass on making a card for this one. |
| 思考 | Once again, 8 out of 13 appearances from Sherlock Holmes. I’m starting to think I don’t need to know 考’s on reading all that much… But I do know 思, yet often struggle to remember its on reading, so this one’s a perfect candidate for a card. |
| 考古 | Related to my forgotten burned card, this one appears thanks to Sakura’s father in Cardcaptor Sakura. I should know the pronunciation of 古 (burn almost two years ago), but I feel like I’ve never seen this kanji before, so this should make a good card. |
I do have more anime subtitles I haven’t parsed yet. If I had, some of these words may have had higher frequency (outside of Sherlock Holmes).
For this, I let Migaku’s kanji add-on to all the work. It creates a kanji card, which also shows common words that use that kanji, as well as words from my cards that also use the kanji.
For kanji card reviews, my main objective is to learn the “keyword” for the kanji. And for that, I have a lot to choose from:
| Source | Keyword |
|---|---|
| WaniKani | think |
| Heisig RTK | consider |
| Kodansha KLC | think |
| Tuttle Kanji | consider |
I’ll go with “think”, as that’s what comes to mind for me when I see 考.
I grade sentence cards one of two ways:
If I should know all the kanji, then I judge on reading the whole word, as well as its meaning.
If there is a kanji I don’t know, I’ll check the furigana if needed. I’ll judge on reading the kanji I should know, and the meaning.
The number of cards has been manageable so far, but I know it’ll be too many for me soon enough.
I have two plans for when this happens:
Suspend cards that I know are really easy for me.
Use Migaku’s retirement add-on to retire cards that I’ve gotten right enough times that my next review won’t be for a long time (such as over a year).
The latter one won’t be usable any time soon since I’ve been using Anki for this only a short time. Currently, of my 111 vocabulary cards in this deck, the furthest out review is for 15 days from now.
Today’s discovery:
Recently re-learning 相 (lv 9), 想 (lv 13), and 箱 (lv 16) together has made it easier for me to distinguish the three.
It’s too soon to tell if this will last, but I’ve always had trouble with 相 (unlocked 13 months ago, currently enlightened) and 想 (unlocked 9 months ago, currently guru). At the moment, going over Anki cards for vocabulary like 真相 and 相談, alongside cards for vocabulary such as 予想 and 回想, is making it more readily clear the difference between the two.
But maybe after a few more days (or a week), I’ll be confusing the two again?
As for 箱, I always recognized it as “that kanji with 竹 on top, and something of a boxy shape”, which allowed me to burn it in a mere seven months. By the time I learned other “kanji with 竹 on top, and something of a boxy shape”, 箱 was far enough that I knew “the kanji with 竹 on top, and something of a boxy shape and is up for its enlightened review” was box. Hopefully I’ll learn it properly this time!
It’s been about three weeks since I started auto-passing most of my leeches in Apprentice. How has that changed things?
My total Apprentice used hover around 100. Now it hovers around 80.
My daily reviews used to about about 100 to 160 cards per day. Now it’s around 50 to 100 per day (not counting auto-passed Apprentice leeches).
I’ll admit, I was expecting a bigger drop than that.
I’m still not ready to resume level 29 lessons. I’ve just about finished guru’ing all the pre-level 29 cards that were added in recent updates.
I’ve started improving my Anki setup a bit. My current decks are:
I think I’m on just about on the edge of my Anki kanji+vocabulary reviews going from “these are easy to remember” to “why can’t I remember this one, either?”
