How is Wanikani choosing which words to teach?

When I did new lessons today, in the example sentences of one vocab I came across the word 取引(とりひき, deal). I knew the word from my grammar studies, I knew both the Kanji that made it up because Wanikani had already taught them to me. Yet I couldn’t read the word. Because I knew it was a more or less common word that comes up once in a while, I checked whether or not Wanikani was teaching it. Turns out, no, they decide not to. And then like two vocabulary lessons later, 汽笛(きてき)pops up, meaning steam whistle. Honestly, I don’t think I used that word even once in English or German, so why is this one being taught instead?

My problem with this is as soon as I make the jump to immersion and start reading native Japanese material, words like 取引 that I should know but can’t read are going to pop up. For every one of these, I’ll have to look them up and create a vocabulary card. If Wanikani refuses to teach a word like that, how many more of these words am I going to encounter?

I fear that even if you do switch to native material after reaching level 60, you’re still not going to be able to read at all. And that’s the reason for starting this whole thing in the first place.

7 Likes

It can also be written as 取り引き - the hiragana just condenses into the kanji. This happens a lot for words like these

10 Likes

In general they seem to choose words to give you multiple exposures (where possible) to the common readings for each kanji. There is some preference for more common words, but there are plenty of common words that are not in WK, especially ones that use common readings that are already covered in other vocabulary words.

Where I’m at, I find that I recognize a fairly substantial chunk of words when reading native material (and I would say my kana vocab is more limiting than my kanji vocab), but there are definitely words that have known kanji that I don’t have the right reading for. On the plus side, I know at least one, often a few words using that kanji and can usually retrieve one to be able to type the word in to check the meaning and reading in a dictionary. Whenever you switch to native material, there will be things that you can’t read, with uncommon readings, exceptions, less common kanji etc - you can expect to still look things up (and make more cards if that’s your preference for learning new vocab - personally if I know the word without kanji, and I know the kanji, I usually find it not difficult to link the two and don’t make new flash cards)

11 Likes

For each kanji they try to give at least a few vocabs and try to cover all the readings. So for 笛 there are three vocabs, and the only one with onyomi てき is 汽笛 because it’s probably the most common word using this fairly uncommon reading. Also 汽 is taught just one level before so it works neatly.

The problem with 引 and 取 is that they are such high frequency kanji that there are literally hundred of words involving then. Which one WK should choose ? There are already 15 vocabs with 取 and 10 with 引. Now, 取引 is really common so I actually think it would be a good idea to have it in WK, but there are many other common words involving 引 and 取 not in WK…

13 Likes

Others have covered this but yeah, for what it’s worth, that word is labelled common on Jisho, and at a glance it looks like the only one tagged common that has that reading.

The unfortunate news I have to break to you is, because Wanikani doesn’t teach literally 10s of thousands of words… the answer is 10s of thousands. This isn’t Wanikani’s fault for the words they’ve chosen, there are just so many words. At level 36 and a pile nearing 4000 words I’ve mined on my own, even things I can read comfortably (with a dictionary) feature a new word every couple sentences at minimum.

17 Likes

Others have already touched on Wanikani’s reasoning here, and covered the important parts of that, but this line throws up a flag for me, so I want to offer some advice:

Do not wait until level 60 to switch to native material.

You absolutely need to be doing reading and immersion well before that. Graded readers, something like Satori Reader, and manga/light novels with furigana (depending on your tolerance level for frustration and ambiguity when you start) are a good way to start doing that. Join the book clubs here on the forums. Get exposure to the language. If you put off starting to read native material until you think you can read every single kanji, you are in for a really massive disappointment on a couple of fronts:

First, Wanikani doesn’t teach every kanji. They teach a lot of kanji, and a lot of the most common ones, but there are still plenty you’ll not have seen if you are relying solely on WK for your kanji learning.

Second, you’ll have spent hours on flashcards, and start to realize that after you have spent all this time on the flashcards, you haven’t been using the words, so your memory of them is likely going to be somewhat iffy, or couched in the (sometimes not great) English gloss that WK gives, instead of actually having seen the word used in native material and figuring out the nuances that surround the word.

You’ll really be hamstringing yourself if you put off native material that long. At level 19, I would recommend you start right around now, frankly. I joined up with the ABBC when I was mid-20s ish, and I still feel like I put it off a bit longer than I should have.

Maybe this comment was one for effect (I could go through the whole program and still not be able to read!), but I just want to be sure you aren’t making the mistake of thinking you should wait to finish WK before you start reading. The important thing here is WK is a tool to help jumpstart kanji reading. It does not teach you to read Japanese on its own. You have to do that with grammar studies and immersion. WK just helps to smooth the process.

26 Likes

A while ago I played with the BCCWJ corpus and made a list of 100 kanjis words (filtering out kana words) that are not in WK but are in the 1000 most common words according to the BCCWJ.

I didn’t clean up the list so there are some skippable one, like some numbers in kanji at the beginning, or thing like 大きい vs 大きな, but anyway:

100 kanjis word not in WK but top 1000 frequency

二十
三十
九百
六十
九十
八十
事業
七十
二千
知れる
企業
利用
年度
大きな
調査
感ずる
八百
平成
規定
実施
地方
中心
多く
整備
願う
我が
全体
生産
機能
箇月
当時
図る
向かう
自身
経営
年間
事務
六百
推進
七百
発生
確か
保護
表示
設定
定める
今後
市場
一部
関わる
団体
御前
三千
基づく
行政
代表
検討
市民
進める
指定
小さな
十分
明らか
思い
法人
作成
年金
取り引き
国家
当該
現われる
連絡
具体
同様
五千
通信
申し込み
生ずる
業務
金融
様子
成長
高齢
特定
設置
放送
通ずる
安定
一方
信ずる
参考
発展
運転
開催
申し上げる
確保
参照
立場
介護
改革
調整
表わす
減少
回答
体制
週間
所得
応ずる
実現
問い合わせ
認識
最大
学習
金額
作り
含める
目指す
資産
分野
因み
国内
次第
課題
本人

Lot of business words, so probably the corpus have many newspapers and such. 取り引き is in there btw !

11 Likes

Thank you very much for calling me out XD

4 Likes

We need an Anime and Manga Corpus.

2 Likes

I actually came across this in a childrens book recently and I cannot tell you how delighted I was.

I’ve just been through a couple of levels with a lot of baseball terminology and yes I know it’s popular in Japan but for those of us outside the US, it really doesn’t help with learning the kanji because I have no idea what those words mean in english (I added ‘baseball term’ as a synonym because if I ever come across them in a japanese text I reckon that will suffice for my understanding).

5 Likes

I know nothing of baseball. I think I’ll follow your lead and just put “BT” as my code word XD

1 Like

That’s correct. You’ll still have to learn a ton of vocab and grammar. You don’t have to make a vocab card for every new word if you don’t want to, though. Just look it up and move on.

As others have said, you should definitely start working on native material earlier than that. Wanikani (kanji) is just part of the equation.

8 Likes

Here’s a link to the Absolute Beginner’s Book Club Home Thread. They are currently a good ways into the manga ハピネス. The next book on the schedule is a collection of stories about/from each of the Japanese prefectures, I believe. :grin:

The Beginner’s Book Club also just started with the first chapter of Yoru Cafe this week, if that catches your interest a little bit more. Just in case you choose to join a book club to get started reading. :grin:

4 Likes

I’m not expecting Wanikani to teach me vocabulary, I’m expecting it to teach me how to read. Naturally, encountering unknown words is inevitable when switching to native material. Even in English or German I still encounter words that I have to look up.

The thing that bothers me is that I cannot even read words I know, even though they might be fairly common. It would be rather annoying if most of the words I end up looking up are words I already know, and the words I don’t know come on top of all that.

My wording on that sentence might’ve been at fault for this one. I’m not planning to hold off on immersion until Level 60. What I was trying to say is that even if you were to hold off for that long, you’d still barely be able to read.

I don’t think your Wanikani level is telling of when you’re supposed to start at all. You could have every item on Wanikani burned and you might still not be ready. The way I see it, what really counts is grammar and vocabulary. Hence, I’m not waiting to reach a specific Wanikani level, I’m waiting until I finish my textbooks. And until I do, I don’t have to worry about getting rusty about what I learnt, because I’m getting at least an hour of exposure every day.

That’s amazing! I feel like if there was an extended list of that, it would really complete the purpose for which I’m using Wanikani. Then you could pick out the ones that would be helpful (obviously, vocabs in there like the numbers you’d just skip) and study those. That would be so much more efficient than being perplexed stumbling upon every single one of them in native material.

Is there perhaps a list of top 10000 (or 5000) most used Kanji expressions? Is there a way to immediately compare that with Wanikani’s vocabulary database and remove every matching word?

2 Likes

I really don’t understand this distinction you’re making, sorry. The way words are written and read with kanji usually have a coherent traceable history if you dig into them fully, but from a modern perspective, they’re fairly arbitrary. The words that you know how to read without having already read them are effectively zero. I mean absolutely if you use WK’s common readings and apply them you’ll guess right frequently enough. That would be a little hard with this particular word because as someone pointed out, the okurigana is kind of inconsistent.

If you’re not looking for Wanikani to teach you vocab but want it to have taught you how to read this word… I don’t really understand what you want. It does teach you 取る and 引く which is the closest you’re going to get without having the exact word taught as vocab. But if they did that, your problem would just be another word down the line soon. If you couldn’t read this word then for reading… you kinda didn’t know it. You seem to be defining knowing purely as knowing it phonetically but I guess the misapprehension you’re under then is that kanji is going to be so patterned and logical that you can always make those connections given what WK has taught you. It frequently doesn’t work out that way. There’s a good reason to kickstart your reading by independently learning kanji like this, but in the end the thing that matters is just the words. And the only surefire solution for every word you haven’t encountered in written form specifically, is to look it up.

15 Likes

Yeah, I don’t understand OP either. Fluent reading is pretty much seeing words you know and have seen hundreds of times, not just guessing compound words readings.

Can’t learn how to read without learning the vocab.

11 Likes

Respectfully, this part seems to have gone by:

I agree with this in that if WK is your only study source, you won’t be ready to read no matter what, but I also just use WK level to gauge how long somebody has been at their journey (operating under the assumption that everybody knows WK shouldn’t be the only thing you’re doing). But if you believe this, I’m very confused by your stating that you expect WK to teach you to read… I may be misunderstanding something, but it feels like you’re flipping back and forth… Your WK level has nothing to do with when you should start reading, but by level 60 you should be able to read? (Which, you should be more than ready by that point IF you are learning grammar elsewhere). Those feel like contradictory statements…

Regardless, you should start to read when you feel ready, of course. I can’t tell you how you should study, or guarantee that the way I’ve done it would work for you. Every person is different, after all! :grin:

That being said, anybody who has learned from a textbook and then tried to read native material will tell you that there’s a rude awakening involved there. Textbooks are a sanitized version of the language designed to make introduction easy. Even after finishing a textbook, you aren’t going to be able to read easily, I’m afraid. That’s why I encourage jumping into native material as early as you feasibly can; that point will be different person to person, of course, and dependent on their level of patience with ambiguous phrasings, contractions, grammar points you’ve yet to learn, and the like.

There are core 10k decks, and Tango N-Level decks that exist for Anki, but I’m not really sure where. A Google search for either of those terms will probably get you there. They will include kana only words too, though, I’m sure.

As far as matching it to the database, it probably would be fairly simple to write a program to do just that (somebody may have already done so, honestly), but that would require some searching as well (or the time/ability to write the program yourself).

5 Likes

The reason I didn’t respond to that point was because I agreed with it. I didn’t have anything to say that would contradict it. I felt like it was written based on the assumption that I was relying on WK for grammar and comprehension, which is not the case. So I can’t respond with anything other than “I would have said the same thing”.

When I say I want WK to teach me how to read, I literally mean just that: reading. I can read Norwegian, probably not understand much of it, but I can read it. If I wanted to understand it, I would need to learn the vocabulary and grammar of the language. I’m not putting this responsibility on WK. If there is a sentence I can read in Kana, I want WK to help me read it if it’s in Kanji. If I cannot understand it even in Kana, then it’s the same situation as in any other language.

By level 60, I feel like you should be able to read most of what you see, even if you can’t understand it (except for the Kanji words you learnt through WK). The reason why your WK level isn’t relevant is because there’s no point in reading if you can’t understand anything.

Although you made a fair point that the WK level could be indicative for how long a person has been studying Japanese, which is actually a good guess in my case because I’ve started WK and my grammar studies around the same time. However, I’ve put about 2 to 3 times as much time into grammar.

I think this greatly depends on the textbook itself and how that person used it. Nonetheless, someone who picks up a textbook with the expectation of being able to easily understand anything afterwards will have a hard time facing reality regardless of how effective the book was. Such expectations are not really suitable for a textbook.

Still, a textbook is the more efficient way to learn grammatical structures. Imagine these two cases:

Case A: You encounter a new grammatical structure while immersing yourself in native material. You won’t be able to understand how it works from that one sentence, and you won’t be able to understand that sentence properly. Worst-case scenario you don’t even realize that it’s something new (e.g. passive and potential verbs often share the same form). Depending on its complexity, you might need to see that structure hundreds or thousands of times until you get the gist of its functionality, and you might still get a false idea of how it actually works.

Case B: You encounter it in a textbook, read an explanation of what it means and how it works, do a couple of drills, and then you jump into immersion. You come across that structure and actually understand the sentence, and those thousands of times you encounter it afterwards become repititions to help you internalize it.

A is obviously less efficient, requires a patience that most people can’t be expected to have, and thus makes them very prone to ending up burn-out. And this is why I think textbooks should be put first. They provide so much knowledge that I have yet to learn. I couldn’t imagine attempting to acquire all that knowledge through constant cryptography-like guessing games rather than just sucking up the information perfectly laid out for me to learn methodically.

My stance on this is that textbooks are for studying, and immersion is for internalizing.

(This part of the conversation completely derails the actual topic, but I still wanted to respond because I’m really passionate about this part of language learning.)

4 Likes

That helps a lot to understand what your expectations are. But again, I’m sad to say, Wanikani is not going to do as much as you want it to, for several reasons. One is simply all of the kanji not on WK, when you say you want to be able to read “most of what you see.” I’m reading a visual novel right now that I’d consider on the easier side and it’s actually startling how many non-WK kanji come up every day. But that’s not all.

Kanji have a whole lot of readings, so many that WK doesn’t teach many even for the kanji it does teach because the scale would balloon out to something monstrous. Does a word use straight onyomi readings or not? There are tendencies you can guess with but that’s never 100%. Same for rendaku and other exceptions.

That’s not even getting started on words that can be written in a bunch of different ways. Take あなた, a simple word for “you.” It’s commonly written in kana so Wanikani doesn’t teach it at all. But it’s not always. Sometimes it’s 貴方, and sometimes it’s 貴男. Both of these are essentially “exception” readings. You can’t prepare for reading this word any way other than already reading it, yet when you do, you’ll shake your head at the fact that it’s a word you, in one sense, already knew. I definitely have multiple vocab cards for the same exact word written in different kanji because I have to figure out every way I could see it.

I don’t mean to be argumentative at all! But I think kanji and reading is a larger beast than you realize from what you’re describing. To be fair, with so much inconsistency in pronunciation in English, I’m not sure they’re substantially different in that way. It is what it is. But Wanikani is only the beginning of being able to read, even under the definition of simply being able to figure out the sounds of what you’re looking at. You’ll always be checking on readings.

13 Likes

Well, if the point is to be able to sound out everything, then I think you can do that by level 60. Of course, it depends on what you mean by “most”. You will keep encountering a lot of kanji you have never seen before (that are not taught on WK).
For kanji you know, there are some rare(r) reading that are not taught on WK.
For compound words, roughly 80% follow a on’yomi-on’yomi pattern. However, some kanji have multiple on reading, and rendaku can make things a bit annoying, but you’re still mostly okay.
The next category is kun’yomi-kun’yomi (like 取引). You’ll also get a feel for those eventually, and won’t mind the lack of okurigana (so you’ll be fine with 受付, 締切, 書出し, …)
Then, you have mixes of on and kun (like 台所), and words where the kanji are just there for meaning, with reading completely unrelated (like 山葵 わさび). There’s basically no way to read those if you don’t know the word already.

So, you’ll never be able to just read Japanese. However, depending on the tolerance implied by “most”, level 60 will allow you to reach your goal (a lower level, like 30, might even be enough).

9 Likes