field notes, languages, personal history

中文: the first 1000 characters

Rediscovering the forgotten advantage of rote memorization.

I recently reached the 1000th character in my deck of Chinese characters on Anki, my flashcard app. It seemed like a good occasion for a blog post celebrating the fact that memorization still works. By the end of this post, this pie chart should make sense to you:


A couple of months ago I wrote a book review that’s now sort of “trending” on Goodreads–I happened to review the book the week it came out, and must have been one of the first. My review is currently tied for the most-liked review of that book. The book is Ultralearning by Scott Young, I gave it 2/5, and my review reads, in its entirety: “Just maybe 10x longer than it needed to be.” 😂

Still, it may be worth the money just to skim through Ultralearning and pick out what’s new to you. For me, it was worthwhile to be reminded of a few of the big components of learning. One is ensuring that you understand the big picture of what you’re trying to learn: the whys, causes and effects, overarching narratives, recurring themes, etc, without getting bogged down in details. There are lots of good tactics on how to do this.

But another big component is, at first glance, just the opposite: rote memorization. Drills. The weakness of my schooling, growing up, was that it focused on memorizing trivial things without any of that big picture context or critical thinking. No wonder we grow disillusioned as adults and ditch that whole system.

But more recently, I’ve hit a stage in my language learning project where rote memorization is really my biggest blocker. See, I’m currently working on Mandarin and Cantonese. But having grown up in a Cantonese-speaking household, and often hearing Mandarin spoken on TV (or sung in my Walkman CD player!), I’m in an interesting position as a language learner. I’m already comfortable with the grammar and pronunciation of both dialects/languages. I have everything you need… except for words. I just have a really small vocabulary, and need to learn lots and lots and lots of words.

I also never quite learned to read Chinese–which, to an even greater degree than learning meanings of words (which can be learned in context, like by watching more TV), is a matter of pure rote memorization. I need to learn a character, and know its corresponding sound(s!), including the correct tone(s!), and know its meaning(s!) and what words it’s found in. And just repeat that thousands and thousands more times. Yes, there are some patterns, so it does get progressively easier to learn new ones, but the patterns not very systematic. There’s no real shortcut. There’s also no real excuse for me not to do it, as it’s basically the only thing keeping me from being fluent.

(Digression: Okay, but who came up with this writing system, anyway?! Especially gems like these: 千 干 于 (qiān, gàn, yú, respectively), or these: 未 朱 失 (wèi, zhū, shī) which are the bane of my existence. Yes, Chinese characters hold layers of meaning and history. But the cost of that romanticness is that it’s also taxing as hell to memorize them and tell them apart. They are gorgeous. Gorgeous and annoying.)

Facing the imminent reality of thousands of characters to memorize, I thought back to earlier times when I’d relied on rote memorization and it had served me well. Before college, before high school, before all the failed learning strategies taught in school.. there was 6-year-old rory, trying to memorize the times table. 6 x 8 is 48, 6 x 9 is 54..


In Outliers, Malcolm Gladwell has this theory about why Chinese kids end up better at math. (Caveat, I read this book like 10 years ago, so I may not have it exactly as he wrote it, but this idea stood out to me so much that I still remember it now.) Part of it is that Chinese numbers take slightly less time to say (already a questionable statement), and since the window of short-term memory is just a few seconds long, Chinese kids can fit more numbers into their memory, a slight advantage which compounds over time.

Lol. I suppose it could be a factor, but why do I think that I personally ended up better at math than the next kid? Because when the next kid was trying to learn the multiplication table, they invariably tried once, got maybe halfway, and then left to go play soccer or whatever. I practiced math, completing assignments from my dad, for hours, every day, including all summer. In the period of time when I was learning the multiplication table, I practiced it in both English and Cantonese, repeatedly writing out each row, then reciting the row out loud. Only when I had the 6’s row perfect was I allowed to move on to 7’s.

In case you’re wondering if this was a tiger mom / tiger dad type thing: nah, let’s be real. I had my bad days when I resented the whole thing, but overall, I loved it, even then. No form of learning in life since has been as satisfying as the education my dad gave me in those days. Because beyond multiplication, exponents, and algebra, my dad imparted to me a general principle for me to carry through my life: that the ability to use your own mind, to endlessly expand what’s in it and then call upon it when needed, that ability matters. And the next kid’s parents imparted to them the principle that it’s not really going to matter, because in the real grownup world we’ll all use calculators anyway.

The absolute best days at school were whenever we had a timed test for math. I’m not kidding. 3 minutes to complete 100 problems (where each problem was like “7 x 3”)… it was the performance of what you’d practiced, pure and simple, no bullshitting through it. You don’t get anything for finishing first–just glory. But I always aimed to finish first with 100/100. And I often did, except the year that I was in the same class as the new kid, Tyler, where we’d kind of alternate being first. No one else came close. Naturally I was deeply in love with him.

(A note of interest: it was common for tests to go up to 10×10, but I deliberately chose to memorize up to 12×12, because I was that kid–and I actually feel that being able to multiply 12’s is one of those little things that has paid off disproportionately well throughout my life, compared to the investment up front. So if you have kids or something, maybe have them go up to 12.)

So anyway, I’m disappointed that the same Malcolm Gladwell who also popularized the 10,000 hours idea (work > talent) would neglect a far more likely explanation for high performance–that kids who put in the work, and kids who believe that it matters, perform better–in favor of some fancy cognitive-linguistic storytelling. I feel like people who endlessly pontificate on the whole question of Asian kids and math are often turning a blind eye to the raw number of hours of focused work involved. In fact, once you get into the habit of seeing, in these discussions, the subtext of white Americans finding creative ways to rationalize away the idea of another ethnicity performing categorically better than them at something–rather than maybe trying to learn from it–you can’t unsee it. “It’s just ’cause X makes it way easier for them”; “but at least they’re not as good as us at Y”; etc. Reader, don’t be that person.

The point of this brief visit back to the 90s: Lots of things in life are less work than they seem. But some things are exactly as much work as they seem. Believe in your brain, believe you can put a lot of new stuff into it if you put in the work, believe that it matters, believe you can retain and recall all of it, more than you ever thought possible. If fewer and fewer people these days believe in memorization (as seems to be the case), that just means that whoever is using it will have that much more of an advantage. You’ll have your answer instantaneously, and it will be correct, while the next person is still reaching for their phone to open Google Translate or the calculator app, because they’ve allowed their beautiful mind to wither away.


That said, even rote memorization can be done far, far more efficiently than people have conventionally done it. That’s where spaced repetition systems (SRS) and Anki come in. I won’t go into detail about how SRS works, as there are tons of great explanations on the internet, including this one in the intro to the Anki docs. Basically, it’s a way of timing your flashcard reviews so as to maximize retention.

Anki is the SRS-based flashcard app that I use. Currently I’m using it just to memorize Chinese characters. I imported a deck of the 3000 most frequent Chinese characters and have been working my way through it for the past couple of months.

A really nice thing about learning Chinese is the existence of HSK, a widely used proficiency test that effectively standardizes the proficiency levels, along with the vocabulary you need to know at each level. Thousands of Chinese characters have an HSK rank (1 being the most basic/frequent, up to like 5000 or something). My flashcards include the HSK rank of each character, so I know approximately how advanced it is.

I want to point out that I’m not recommending this as the best way for anyone else to learn Chinese. Heck, it’s not even the best way for ME to learn Chinese. Most units of meaning in Chinese (beyond the basic stuff) occur in groups of 2 or more characters (let’s call them “words”). For example, 电脑, computer, which is a pair of characters: literally “electric brain.” I think it would be way more effective if I used a method that included words, and not only individual characters. But I’m lazy–too lazy to look for another deck–and stubborn, and I already have this deck, so I’m going to brute-force my way through it.

Now we can actually revisit the pie chart and what it means.

It’s a chart of where I’m at with all 3000 cards in the deck. “Unseen: 2000” means I’ve seen the other 1000 of the cards at least once. It doesn’t mean that I know all 1000 of those; I’m not yet able to consistently recognize all the cards that I’ve seen. “Young+Learn” are cards that I’m not yet doing consistently well with. “Mature” are cards I’ve been doing consistently well with for some period of time. “Buried+Suspended” is some obscure Anki stuff I won’t go into here.

I have it set so that each day, I get 20 new cards, and a bunch more old cards are up for review, up to a max of 200. It takes about 20 minutes a day total. The more recently I learned a card, the more likely I am to get it wrong in review, but the more likely it is to come up for review. Which means I get things wrong quite a lot, and some days (a lot of days) I start to feel really dumb.

BUT! I also already have 388 “Mature” cards. Meaning, I’ve already more or less memorized 388 characters of Chinese! It sounds like a lot, when you put it that way. It’s really comforting and also mindblowing to know that over time, if I just keep doing my 20 minutes of reviews each day, that “Mature” section of the pie chart will grow and grow, until it reaches 100%. And then I really will have to find another deck–or just switch to another way of learning new characters and words.

The little I’ve learned so far has allowed me to start texting my parents and friends/family in Chinese, which is one of those things I’d never thought would really happen in this lifetime.

So: Believe in your brain. And a little algorithm that helps your brain do its best work.

Further reading:

The book I was talking about, Ultralearning, also has a lot of good tactics for memorization. The entire chapter on retrieval is really important, especially because it’s really counter to how everyone (at least in America) thinks about effective studying. It’s just funny to me that our standards for how to learn have fallen so low that a book about basic, fairly uncontroversial learning practices gets titled “Ultralearning” and gets all this marketing hype around it. Like, the book should be called just “Learning.” 😂

Want to Remember Everything You’ll Ever Learn? Surrender to This Algorithm — nice lengthy backstory on how SRS was invented.

You Don’t Have A Foreign Language Problem, You Have An Adult Literacy Problem — a rant about how the Western world developed the mindset that Chinese/Japanese writing systems are hard, and how one might go about reversing that mindset.