Note on why this person is taking an unusual route: from https://blog.kevinzwu.com/cyborg-learning/ , they are a "second generation Chinese immigrant" and "heritage speaker"; that is, they live outside China, can speak the language because they learned it from their parents, but cannot read it.
Edit addendum: https://blog.kevinzwu.com/chinese-cursed-logographic-dags/ is a fun read. I've been using the imaginatively titled "kanji study" app, which uses the same Outlier database mentioned which has the graph based etymology.
There's an additional level of chaos when learning the "same" characters as kanji rather than hanzi.
I liked the 10% @@@ example, demonstrated their point pretty well.
Also for anyone who speaks or is currently learning Chinese... I've been working on a multiplayer CJK word game that shares a similar efficient brute force style of learning to the author's approach (although presented via gameplay instead of tooling). Every turn you get a random character and must type in a word that contains the char in ANY position. If you like fast paced word games it might be up your alley: https://danobang.com/?game_lang=cmn
> I would end up copy-pasting interesting words into the dictionary window to pull up the word entry. SLOW!
> I would then click on the component characters to open their nested dictionary entries. SLOW!
> If I needed to remember the stroke order, I would scroll down for the static display. SLOW!
So, all of these are included in Anki-xiehanzi(https://github.com/krmanik/Anki-xiehanzi). Free open source software like Anki & xiehanzi can save you from using all those tokens.
It is also allowing me to see all relevant associations easily when revealing the card in built in SRS, you add cards to SRS as you browse, so they are related to what you already know / currently exploring.
Mind you, all data visible is collected from different reputable available sources. When you click "explain" there's a clearly marked LLM explanation, but my explanation generation pipeline pushed all generated explanations through 5 different models including all top Chinese-first for verification, and on average it took a few iterations back and forth to iron out any information that could potentially mislead the learner.
this looks incredible and exactly like something i've been wanting. is there the same amount of depth for the 9k+ characters? if this is open source, id love to build on it;i was wandering if op had posted his on github.
Only about 5K explanations now I'm still trying to polish the pipeline before covering more. Due to all these verification and associated regenerations the cost is quite high.
It's not opon source but it is completely free. Open sourcing is on the table but currently it would be additional work and distraction. Just licenses seem like a headache, from a quick poke even when ok for use on non-commercial data redistributing may not be ok. So not any time soon, but if you are working on something similar I can at least share detailed datasources, finding good ones was not easy and LLM integrate them fast.
Mostly datasets but who knows if LLMs stuff also if I'm creating a datasets with them[0]. Long story short it's a potential headache, don't want that headache for now. Plus basically two options after open sourcing: people are not using it (time and effort wasted) or people are using it and then it's a chore. But still on the table. But I'm currently not close to the table.
> I opened Claude Code and started rambling into my mic. It wrote thousands of lines of questionably efficient JavaScript. I didn't read a single one.
Hm. I always knew voice mode was a thing, but I have never tried it. What's people's experience with it?
Being able to correct my words is a good thing. Hell, I did it ~3 times when writing this comment. I can't do that when I'm rambling. I'll trip, or CC will think I'm finished.
I maintain an Anki deck for my chinese learning. Following the HSK books, I add new words to my deck with the character on front side and pinyin + definition + audio (from the CD and sliced using Audacity) on back side.
WIP (need more work in multi-hanzi words), but won't stay in the same 5 words for more than a day. it has been working well for me
the most interesting thing was GPT helped with the sentences and simplified words meaning and bing translate provided the audios
the goal is get the ~2000 words you need to be proficient in 1 year, 5 words a day plus refresh old words, also it keep track of your progress against the year, no streaks
>Empty landing page with a "Sign in with Google" button
>Can't find anyone else talking about it online, no screenshots of the gameplay, nothing
That's gonna be a no from me, dawg. It sounds like a cool idea, but sites have to get better about asking you to hand over your credentials without even telling you what exactly your getting for them.
I very much agreed to the "build your own tools for learning" route they took. I made my own site to practice Japanese grammar as it was hard to get feedback on that. tokidoki.meertens.dev -> it's open source so anyone can adapt it to their needs as well!
I can speak Chinese fluently but I need to improve my reading hugely. This sounds like exactly my usecase. I would actually be willing to pay for this (though less certainly for a subscription, preferably Pleco-like one-time fee - even if larger).
Interesting process. I wonder if he considered doing this with Anki. That would have given him a good SRS algo for free and Anki cards are also HTML+CSS+JS. I probably wouldn't try to put LLM calls onto my cards though
> I decided to go against the grain of the near-universal advice to "learn to read by reading".
...Why? That advice is universal for a reason. The side adventure with Claude Code strikes me as a distraction from the fact that there is a hard thing you want to do but are avoiding because it's hard.
This is a hilariously common thing with studiers of Asian languages. There are countless posts with people spending years, even more than a decade, just trying to memorize every single kanji and how to write it before even beginning vocabulary or basic grammar, then lamenting how difficult the language is and how they can't pass kindergarten level tests. So then they spend loads of money on apps, make custom tools, and find countless other ways to burn time.
Meanwhile others read books and get pretty good at their language of choice in a couple years.
Lived in Taiwan for years while studying traditional Chinese. This is spot-on. There’s this bizarre almost Pokémon level obsession (gotta catch'em all!) among foreign language learners with a STEM background where they fixate on amassing huge numbers of “memorized words.” Learning words in isolation is exactly what you don’t want to do, you need to see them in their context.
It’s like thinking you can get good at chess by just memorizing how each piece moves. You need the board, the surrounding "context", and not just study in isolation.
I agree completely. I will focus of Japanese because it’s the language I have experience with.
With a good order (RTK), optimal reviews (SRS) and putting 30 minutes a day it is possible to learn keyword to Kanji writing in a couple of months to one year. Make it two if you are a busy person. After that you need 10 minutes to maintain the knowledge per day. (I’m assuming 2200 Kanji).
People that did that successfully will recommend it to be done as early as possible as they know the boost in learning it provides.
I think it’s a trap, because it’s possible to get to a very useful level in the language while ignoring Kanji, and most people will be perfectly happy staying there. At that point you will have a much better idea if you really need to go all the way.
I'm at HSK3 level and struggle to find things to read outside of my actual textbooks with precisely-calibrated texts. If I can't read am average billboard, what should I read to improve?
I'm at a similar level, maybe a little behind that. I don't have any advice for you, but I'll relate the path I am planning to take. Would be happy to hear others' thoughts, too.
My feeling is this level is just too early to read "real" texts, so I am continuing to just use graded readers. I use the Du Chinese app for this, it contains a bunch of short stories at different comprehension levels, and has a spoken accompaniment to each story read by a real speaker (not AI/TTS). I also have some physical books from LingLing Mandarin, I like the challenge of not having a dictionary immediately to hand like I do in the app. My hope is by the time I finish with the Advanced stages of each of these sets of readers, I will be able to start reading "real" texts and fill in gaps with a dictionary app, at which point there is an infinite supply of material.
I do worry I'll end up at the "10% missing comprehension" described in the article, though, at which point I guess I'll try to find even higher level graded readers, if they exist. We'll see.
3. Count how often each word appears and sort sentences by descending frequency of the least common word.
4. Use binary search to find a location in the sorted collection of sentences where the difficulty feels about right.
Of course this gives you a collection of disjointed sentences, but you can always go to the original file and look at the surrounding context when you find an interesting or confusing one.
It is normal that there's no native content at HSK3 level. It would be like looking for native content at A1 level in English. But I found the "actual textbooks with precisely-calibrated texts" / graded readers are not that bad. My teacher is a martian was OK, I want to read The Monkey’s Paw now.
Read simple novels and look things up in a dictionary when you see a bunch of words that don't make sense. Watch shows with subtitles and do the same. Chat with people in the language. This is how non-natives learn English outside of school. It's also how non-natives learn Chinese.
Ain't nobody downloading stuff from GitHub and splitting things into tokens to learn English like the other commenter is suggesting lmfao. But people are obsessed with bizarre and completely inhuman methods like that when they learn Asian languages. I cannot understand why, and it's seriously just Asian languages.
Indeed, his premise is quite flawed. Yes, you will have difficulty understanding everything you read at an early stage, but you aren't supposed to be able to understand everything. You read to heavily reinforce the most common words and patterns that show up constantly, and from that base you pick up bits and pieces as you go along.
Under normal circumstances, even if you grind out "knowing" all the words in advance you will still struggle to read any basic sentence and you've essentially wasted time because it's an unskippable step; to be good at reading you need to read a lot. He seems to already know Chinese though, so this might work for him since he is not really having to learn the language or specific vocabulary, just how it's actually written.
Note on why this person is taking an unusual route: from https://blog.kevinzwu.com/cyborg-learning/ , they are a "second generation Chinese immigrant" and "heritage speaker"; that is, they live outside China, can speak the language because they learned it from their parents, but cannot read it.
Edit addendum: https://blog.kevinzwu.com/chinese-cursed-logographic-dags/ is a fun read. I've been using the imaginatively titled "kanji study" app, which uses the same Outlier database mentioned which has the graph based etymology.
There's an additional level of chaos when learning the "same" characters as kanji rather than hanzi.
I liked the 10% @@@ example, demonstrated their point pretty well.
Also for anyone who speaks or is currently learning Chinese... I've been working on a multiplayer CJK word game that shares a similar efficient brute force style of learning to the author's approach (although presented via gameplay instead of tooling). Every turn you get a random character and must type in a word that contains the char in ANY position. If you like fast paced word games it might be up your alley: https://danobang.com/?game_lang=cmn
Hahahhaha you are everywhere!
I remember learning about your game in a mandarin forum. Thanks for making it, it's very cool!
haha nice to meet you here too, thanks for enjoying it!
> I would end up copy-pasting interesting words into the dictionary window to pull up the word entry. SLOW!
> I would then click on the component characters to open their nested dictionary entries. SLOW!
> If I needed to remember the stroke order, I would scroll down for the static display. SLOW!
So, all of these are included in Anki-xiehanzi(https://github.com/krmanik/Anki-xiehanzi). Free open source software like Anki & xiehanzi can save you from using all those tokens.
Great minds think alike :D https://hanzirama.com/character/%E5%AD%A6
It is also allowing me to see all relevant associations easily when revealing the card in built in SRS, you add cards to SRS as you browse, so they are related to what you already know / currently exploring.
Mind you, all data visible is collected from different reputable available sources. When you click "explain" there's a clearly marked LLM explanation, but my explanation generation pipeline pushed all generated explanations through 5 different models including all top Chinese-first for verification, and on average it took a few iterations back and forth to iron out any information that could potentially mislead the learner.
You can actually see thousands of words I typed just working on that pipeline here https://hanzirama.com/making-of
this looks incredible and exactly like something i've been wanting. is there the same amount of depth for the 9k+ characters? if this is open source, id love to build on it;i was wandering if op had posted his on github.
Only about 5K explanations now I'm still trying to polish the pipeline before covering more. Due to all these verification and associated regenerations the cost is quite high.
It's not opon source but it is completely free. Open sourcing is on the table but currently it would be additional work and distraction. Just licenses seem like a headache, from a quick poke even when ok for use on non-commercial data redistributing may not be ok. So not any time soon, but if you are working on something similar I can at least share detailed datasources, finding good ones was not easy and LLM integrate them fast.
> even when ok for use on non-commercial data redistributing may not be ok.
What do you mean by that? Is this relating to the LLMs or the datasets used as inputs or something else?
Mostly datasets but who knows if LLMs stuff also if I'm creating a datasets with them[0]. Long story short it's a potential headache, don't want that headache for now. Plus basically two options after open sourcing: people are not using it (time and effort wasted) or people are using it and then it's a chore. But still on the table. But I'm currently not close to the table.
0. https://claude.ai/share/23d27617-7608-4c58-93b9-b9613bc29318
> I opened Claude Code and started rambling into my mic. It wrote thousands of lines of questionably efficient JavaScript. I didn't read a single one.
Hm. I always knew voice mode was a thing, but I have never tried it. What's people's experience with it?
Being able to correct my words is a good thing. Hell, I did it ~3 times when writing this comment. I can't do that when I'm rambling. I'll trip, or CC will think I'm finished.
I’ve switched from Duolingo to Immersive Chinese: https://apps.apple.com/us/app/immersive-chinese/id1451785130 It literally starts with wǒ and builds on top of the previously learned vocab
Anything similar to that for Japanese?
> A guy on a forum had hired a calligrapher to write three thousand characters in ballpoint pen
A shame that this amazing resource is not linked.
here you go: https://www.chinese-forums.com/forums/topic/61471-mega-manda...
i’ve used this for a brief while, but dropped practicing handwriting completely shortly after.
I maintain an Anki deck for my chinese learning. Following the HSK books, I add new words to my deck with the character on front side and pinyin + definition + audio (from the CD and sliced using Audacity) on back side.
I'm trying something like Duolingo mixed with Dark souls
https://dondeng.com
WIP (need more work in multi-hanzi words), but won't stay in the same 5 words for more than a day. it has been working well for me
the most interesting thing was GPT helped with the sentences and simplified words meaning and bing translate provided the audios
the goal is get the ~2000 words you need to be proficient in 1 year, 5 words a day plus refresh old words, also it keep track of your progress against the year, no streaks
empty? there's one promise: Learn Mandarin in one year. One lesson a day, 365 days.
that's exactly what it does. free. no ads.
yeah, google it. nobody in there, you won't enter, well you're not a maverick in curve of adoption. that's fine.
hand over your credentials? what's credentials? a Google oauth login?
uh?
I very much agreed to the "build your own tools for learning" route they took. I made my own site to practice Japanese grammar as it was hard to get feedback on that. tokidoki.meertens.dev -> it's open source so anyone can adapt it to their needs as well!
I can speak Chinese fluently but I need to improve my reading hugely. This sounds like exactly my usecase. I would actually be willing to pay for this (though less certainly for a subscription, preferably Pleco-like one-time fee - even if larger).
Interesting process. I wonder if he considered doing this with Anki. That would have given him a good SRS algo for free and Anki cards are also HTML+CSS+JS. I probably wouldn't try to put LLM calls onto my cards though
Awesome journey...I made a puzzle to take a twist of learning the character. Enjoy :)
https://wq-landing.netlify.app/
If it's anything like learning Kanji in Japanese, you basically have to write it repetitively hundreds of times for each word to make it stick.
> I decided to go against the grain of the near-universal advice to "learn to read by reading".
...Why? That advice is universal for a reason. The side adventure with Claude Code strikes me as a distraction from the fact that there is a hard thing you want to do but are avoiding because it's hard.
This is a hilariously common thing with studiers of Asian languages. There are countless posts with people spending years, even more than a decade, just trying to memorize every single kanji and how to write it before even beginning vocabulary or basic grammar, then lamenting how difficult the language is and how they can't pass kindergarten level tests. So then they spend loads of money on apps, make custom tools, and find countless other ways to burn time.
Meanwhile others read books and get pretty good at their language of choice in a couple years.
https://news.ycombinator.com/item?id=47804903 - they can speak it, but not read it.
Lived in Taiwan for years while studying traditional Chinese. This is spot-on. There’s this bizarre almost Pokémon level obsession (gotta catch'em all!) among foreign language learners with a STEM background where they fixate on amassing huge numbers of “memorized words.” Learning words in isolation is exactly what you don’t want to do, you need to see them in their context.
It’s like thinking you can get good at chess by just memorizing how each piece moves. You need the board, the surrounding "context", and not just study in isolation.
I agree completely. I will focus of Japanese because it’s the language I have experience with.
With a good order (RTK), optimal reviews (SRS) and putting 30 minutes a day it is possible to learn keyword to Kanji writing in a couple of months to one year. Make it two if you are a busy person. After that you need 10 minutes to maintain the knowledge per day. (I’m assuming 2200 Kanji).
People that did that successfully will recommend it to be done as early as possible as they know the boost in learning it provides.
I think it’s a trap, because it’s possible to get to a very useful level in the language while ignoring Kanji, and most people will be perfectly happy staying there. At that point you will have a much better idea if you really need to go all the way.
I'm at HSK3 level and struggle to find things to read outside of my actual textbooks with precisely-calibrated texts. If I can't read am average billboard, what should I read to improve?
I'm at a similar level, maybe a little behind that. I don't have any advice for you, but I'll relate the path I am planning to take. Would be happy to hear others' thoughts, too.
My feeling is this level is just too early to read "real" texts, so I am continuing to just use graded readers. I use the Du Chinese app for this, it contains a bunch of short stories at different comprehension levels, and has a spoken accompaniment to each story read by a real speaker (not AI/TTS). I also have some physical books from LingLing Mandarin, I like the challenge of not having a dictionary immediately to hand like I do in the app. My hope is by the time I finish with the Advanced stages of each of these sets of readers, I will be able to start reading "real" texts and fill in gaps with a dictionary app, at which point there is an infinite supply of material.
I do worry I'll end up at the "10% missing comprehension" described in the article, though, at which point I guess I'll try to find even higher level graded readers, if they exist. We'll see.
1. Take a large collection of text, e.g. from https://opus.nlpl.eu/corpora-search/zh-CN&en
2. Split into sentences and tokenize sentences into words, e.g. using https://github.com/fxsjy/jieba
3. Count how often each word appears and sort sentences by descending frequency of the least common word.
4. Use binary search to find a location in the sorted collection of sentences where the difficulty feels about right.
Of course this gives you a collection of disjointed sentences, but you can always go to the original file and look at the surrounding context when you find an interesting or confusing one.
It is normal that there's no native content at HSK3 level. It would be like looking for native content at A1 level in English. But I found the "actual textbooks with precisely-calibrated texts" / graded readers are not that bad. My teacher is a martian was OK, I want to read The Monkey’s Paw now.
Read simple novels and look things up in a dictionary when you see a bunch of words that don't make sense. Watch shows with subtitles and do the same. Chat with people in the language. This is how non-natives learn English outside of school. It's also how non-natives learn Chinese.
Ain't nobody downloading stuff from GitHub and splitting things into tokens to learn English like the other commenter is suggesting lmfao. But people are obsessed with bizarre and completely inhuman methods like that when they learn Asian languages. I cannot understand why, and it's seriously just Asian languages.
The polyglot og Chaz Freeman is interviewed on the CGTN YouTube channel and has amusing thumbnail sketches on his language learning journey.
Indeed, his premise is quite flawed. Yes, you will have difficulty understanding everything you read at an early stage, but you aren't supposed to be able to understand everything. You read to heavily reinforce the most common words and patterns that show up constantly, and from that base you pick up bits and pieces as you go along.
Under normal circumstances, even if you grind out "knowing" all the words in advance you will still struggle to read any basic sentence and you've essentially wasted time because it's an unskippable step; to be good at reading you need to read a lot. He seems to already know Chinese though, so this might work for him since he is not really having to learn the language or specific vocabulary, just how it's actually written.
Did you read part 3? Doesn’t sound like “avoiding hard things” is really a problem for the author :)
https://blog.kevinzwu.com/symbolhead-syndrome/
The hack Chinese app is pretty slick. I really wish I had this when I was taking Chinese in college.