GoKawiil中文 Literacy Speedrun II: Character Cyclotron
09 Apr, 2026
Part I: Cyborg Learning
A great lie of learning Chinese is that reading in a meaningful sense is possible before an advanced level of study. By December of 2025 I had reached a vocabulary of about 1000 characters. On a strict token decoding basis this meant I had about 90% token coverage of typical text. On a semantic understanding basis - well - what does it mean to miss 10% of tokens?
The main @@@@@ is that one out of every ten words doesn't mean a dropped @@@@@ every now and then. Typically, the most important words get @@@@@, and key passages can @@@@@ new words above the base rate. The secondary @@@@@ is a @@@@@ loop that never closes. @@@@@ acquisition is done via repetitive exposure, but low-frequency words need to be @@@@@ in advance to stick. @@@@@ new @@@@@ inline for the first time breaks the flow of reading. Both @@@@@ can be resolved by sticking to "n+1" @@@@@ of @@@@@ difficulty, but I'd rather read nothing at all than educational @@@@@.
With this in mind, I decided to go against the grain of the near-universal advice to "learn to read by reading". The goal was 99% coverage, or die trying. I was going to stop reading, and shove the symbols into my head by force, like cornmeal tunneled down a foie gras duck's throat to fatten its diseased liver.
First, I would need a better feeder.
Observe
The flashcard site I was using was called Hack Chinese. The front side of each flash card was the word, the back its pronunciation and definition. In a separate interface there was a dictionary listing character etymology and stroke order.
I would end up copy-pasting interesting words into the dictionary window to pull up the word entry. SLOW!
... continue reading