Protip: if you keep looking for a new technique to learn kanji, you’re procrastinating.

There’s a common mistake people make when studying in general: they don’t actually practice recall. As long as you try to recall instead of looking at kanji all day and trying to come up with smart techniques or mnemonics or whatever you’ll do just fine.

I don’t understand what is new here, mnemonics sometimes work but if you really want to learn kanji properly you also need to learn stroke order.

In some cases mnemonics work, in others, simply knowing the components works. Sometimes both concepts combine, such as in 木 (tree), 林(woods) and 森(forest). Sometimes it doesn’t work at all, often with kanji compounds, and you just have to memorize it and the various cases for different readings.

Knowing that 悪 (evil, bad, hate) is a weight or i-beam crushing a 心 (heart) only takes you so far. It won’t tell you if it’s aku, o, or waru. You’ve gotta learn it and use it.

> if you really want to learn kanji properly you also need to learn stroke order

i think this depends on what your goals are. for example, if i only wanted to be able to read and type japanese, why would i care about stroke order?

You need proper stroke order to write Kanji with a pen. You want to write Japanese with a pen a lot if you want to learn to read written Japanese because it builds 'muscle memory', i.e. an intimate, up-close-and-personal knowledge of the structure of the components of Kanji. This helps learning to read, a lot. Learning to read Japanese just by staring at printed characters is a lot harder; it's totally doable but it really gets hard beyond a few frequent and graphically salient characters.

If you want your handwriting to look acceptable, then proper stroke order is an absolute must. When you go further and learn cursive and grass script, then you'll get to learn multiple stroke orders per character (and hundreds of conventionalized doodles for components and whole characters).

> Learning to read Japanese just by staring at printed characters is a lot harder; it's totally doable but it really gets hard beyond a few frequent and graphically salient characters.

It might be your personal experience, and I understand that mnemonics don't work out with everyone and some have better muscle memory and worse imagination. However, it's disingenuous to say that writing is a lot harder. I had no problem whatsoever learning meanings of 1500 kanji in 60 days, 2 hours a day. Other people have success too.

> However, it's disingenuous to say that writing is a lot harder.

You quoted me wrong, and frankly 'disingenuous' sounds like an objectively incorrect vocab choice. Apart from that I confess that I speak from experience gained with a small sample size (around 1), and things may differ with each person.

What's more, I assume neither you nor I have done any rigid testing in this matter. 1500 kanji in 60 days does sound a lot, so good for you. However you cannot say with confidence what the future retention rate will be, nor do you (strictly speaking) know how you would have fared with any other approach. Also consider recognition of kanji (and their correct readings) in actual texts and recognition of artistic, calligraphic, handwriting styles; no idea where I'm standing on those but I do feel I have some kind of grip. Training for the latter does include looking at lots of calligraphic samples, and I do repetitively write some of those, too; it's hard to gauge how much either of these activities contribute to the learning effort.

> You quoted me wrong

I'm sorry, you're right about that, and yes, I don't know how good other approaches would have done it for me.

> However you cannot say with confidence what the future retention rate will be

At the time I studied, I stopped doing Japanese whatsoever, so retention dropped as it should. But I think if I have continued reading texts, I would have retained the meanings of kanji, just like it works with other languages I managed to stick with.

You have a fair point, I think we could conclude that both methods have its odds and cons and it's up to each to decide if they want to use it, so long as no rigid testing is provided, as you say.

I agree. There's not much use in writing, and I find muscle memory arguments fairly strange, considering the time it takes to invest to learn this way, as opposed to just do mnemonics and understand the kanji enough. If you're learning to drive, why spend time building cars?

I learned my kanji with Wanikani. Beyond level 30 I found myself ditching the mnemonics and basically "brute force" the learning process. If I had to start all over again I would choose Anki and example sentences (my current learning routine).

I don't think there's really a way around learning by example sentences (either through Anki, reading a lot, ...) so you might as well start from there.

You just have to memorize them. There is no shortcut. Millions of schoolchildren have done this. It's not too hard. Eventually you will start noticing patterns in the characters, which will make it easier, but you will still have to memorize each new character.

> It's not too hard.

It takes 6 years to Japanese children, who are totally immersed in and had prior exposure to this writing system. I think western children are done learning to read in 1 or 2 years. It is freaking hard.

And yet they basically all learn it. Japan’s literacy is very high. It’s not hard, it’s just tedious and time-consuming.

It's not new, and the article seems to state superfluous inventions of their own to generate some income.

It’s frustrating because, on HN, people often miss the key novelty that makes something better and different and unfairly dismiss it. But, in this case, it’s genuinely nothing new.

I find the concept of their 'jars' and the classification into u/ru/i-verbs and na/no-nouns and so on novel, interesting, and also at least not wrong on the intuitive level.

Shameless plug on my own experience with Heisig for Chinese characters - https://underscorehao.net/2020/05/learning-1500-chinese-char...

“We can imagine many little tongues, a 千housand specifically, to help us recall this kanji and how to draw it.”

Well, first of all, my teachers would take issue with “drawing” kanji. You write Chinese characters, just like you write letters in English.

Second of all, my teachers would encourage you to not rely much on mnemonic devices. There is a large body of research about how effective mnemonics are for learning.

From https://journals.sagepub.com/stoken/rbtfl/Z10jaVH/60XQM/full

> On the basis of the literature reviewed above, we rate the keyword mnemonic as low utility. We cannot recommend that the keyword mnemonic be widely adopted. It does show promise for keyword-friendly materials, but it is not highly efficient (in terms of time needed for training and keyword generation), and it may not produce durable learning.

(Note that the paper specifically discusses learning foreign language vocabulary.)

The thing is… it should not take you long to learn Chinese characters. You can just put in the time to memorize it by going over it again and again, and trying to read aloud passages written in Japanese, or answering questions about the content. Really engaging with the material and using it in the same way you intend to use it in the future.

I see a lot mnemonic tricks, apps, etc. for foreign language learning. For some reason, a lot of focus on Japanese, specifically (never seen a post about learning Spanish on Hacker News). My opinion of these apps and techniques is, by and large, very low. If you want to be proficient, you should be reading and writing as much material as possible, and you can remember Chinese characters mostly with rote memorization, with maybe a little bit of mnemonics to help you get started. Only a small percentage of Chinese characters are pictograms. Once you get into the 500-1,000 character range, you’ll be making guesses at character meanings from context and radicals.

Coming up with better mnemonics, or better ways to introduce mnemonics, is a low-value activity. You can go through an entire book of mnemonics that teaches you 1,000 Chinese characters and retain none of it.

What you want is a curriculum that exposes you to kanji incrementally, with plenty of sample text that you can practice reading (and answer questions about!). The sample text should have a high percentage of characters you know from previous lessons, use a few new characters, and occasionally toss in a character (with furigana) that you have to figure out from context.

> I see a lot mnemonic tricks, apps, etc. for foreign language learning. For some reason, a lot of focus on Japanese, specifically (never seen a post about learning Spanish on Hacker News). My opinion of these apps and techniques is, by and large, very low. If you want to be proficient, you should be reading and writing as much material as possible, and you can remember Chinese characters mostly with rote memorization, with maybe a little bit of mnemonics to help you get started. Only a small percentage of Chinese characters are pictograms. Once you get into the 500-1,000 character range, you’ll be making guesses at character meanings from context and radicals.

> Coming up with better mnemonics, or better ways to introduce mnemonics, is a low-value activity. You can go through an entire book of mnemonics that teaches you 1,000 Chinese characters and retain none of it.

That's not my experience at all. Please see my link detailing my own experience with the Heisig method.

> That's not my experience at all. Please see my link detailing my own experience with the Heisig method.

Did you forget to post the link?

Heisig’s approach to break down the characters into components is indeed useful, I just doubt the usefulness of the “stories” that Heisig provides.

Sorry, I posted the link in a top level comment before answering you. In any case here it is again - https://underscorehao.net/2020/05/learning-1500-chinese-char...

His stories weren't really useful to me either and I made my own since that's what the method is based upon. The stories he provides are a guideline for the method, but making your own stories is the key. I've done that over a year ago and I've finished the second tome too since then (albeit with a slower pace than what I did for the first).

How well are you able to read? It looks like the article is about memorizing characters, which is useful, but I’d imagine that the end goal is reading comprehension.

Yes, the end goal is reading comprehension. I'm getting there. Since I've finished the books I've been actively listening/reading while attaining vocabulary. I still have a long way ahead, but not being worried about the characters or how to write them is a big boost.

> For some reason, a lot of focus on Japanese, specifically

That's because Japanese is a popular language to learn, especially in HN-type communities, and to learn Japanese you, most of the time, have to learn Kanji

> never seen a post about learning Spanish on Hacker News

There are some https://hn.algolia.com/?q=learn+spanish. The thing is that you don't have to learn 2000 symbols to learn Spanish. I also think that many "hacker type" people consider raw memorization "boring" compared to conversation or playing around with the language.

> The thing is that you don't have to learn 2000 symbols to learn Spanish.

You have to memorize words, though.

That's true, but it's easier to do while using them. You can "read" Spanish without knowing Spanish. You can't do the same with Japanese.

You can pronounce written Spanish words without knowing what they mean. However, I imagine that most people actually want to know what they are reading.

Also true, although you could get the pronounciation very wrong. It happens to be a few time with English. Regardless, being able to read it make it easier. For example, search for "industria" is easier than searching for "業界", especially for text on a picture. Google translate's OCR is not great, and reproducing the drawings to get a translation is painful and takes a lot of time.

I agree that memorizing words or kanji in isolation from the target language material is not an optimal strategy, whether one uses mnemonics or not. It is not relevant to your argument against using mnemonics.

In addition, in my opinion you understate the usefulness of mnemonics to assist in learning. If we're to agree they do help with learning faster, you might still object as you wrote: "You can go through an entire book of mnemonics that teaches you 1,000 Chinese characters and retain none of it.". It's not much of a concern if one uses spaced repetition software like Anki.

Does Anki actually work if you want to read Japanese?

Flash cards are a really small part of the picture, and making a better system for flashcards isn't going to result in much improvement.

How does this differ from existing stuff like Wanikani?

> I propose a method called and orthographic gradient

typo