Re: Learning Japanese from zero by listening
Posted: Sat May 14, 2022 8:53 pm
(crossposted from my Wanikani log)
A Year of Wanikani Reviews
Since starting Wanikani (again) a year ago, I’ve done a total of 32550 reviews and spent nearly 154 hours on Wanikani. Of those, I passed 24032 reviews, for an overall average accuracy rate of 73.831%. However, that average conceals important variation.
Here is a graph of my (smoothed average) accuracy rate over time:
As you can see, my accuracy was very poor, back when I first started a year ago. It makes sense since I had tons of high level items awaiting review that I never learned properly in the first place, and I hadn’t touched WK in four months, so I barely remember most of the low level stuff anyway.
39 days in, I was forced to give up and reset from level 60 to 48, leading to an immediate jump in my accuracy rate, since I didn’t have to deal with so many of the high level items I never learned in the first place. Note that this graph is a smoothed average, and thus understates the jump in accuracy.
Soon afterwards, I reset to level 47, and then 46, which helped a bit more. However, even ignoring the resets in the first two months, my accuracy has since very slowly climbed over time. It makes sense, since when I keep missing reviews over and over, I’ll eventually start remembering them sometimes, and so I’ve gotten a lot more familiar with the items on WK over the last year. (And getting better at Japanese in general probably helps slightly as well.)
Accuracy wasn’t the only major improvement over the last year either. Here’s a graph of my (smoothed) average review speed:
As you can see, when I first started out, reviews took forever because I couldn’t remember anything. I was only managing 1.5-2 reviews per minute. Even after resetting, I was lucky to even hit 2.5 reviews per minute. Since then however, my average review speed has slowly improved, and I’m now hovering around 5 reviews per minute, and often even better than that.
These two improvements are of course closely related. The biggest thing that slows me down when doing reviews is missing reviews, especially when I check the correct answer or look up related words so I’ll hopefully remember it next time. Familiarity with the words in general helps a lot as well - having more words I know instantly and fewer that I have to agonize about helps me go through reviews more quickly.
Meanwhile, here’s a graph of my (smoothed average) review time spent over the last year:
Counterintuitively, this shows the opposite pattern. In the early months, I spent a lot more time each day doing reviews. (The big dip at the start was when I was on vacation, as are most of the other big dips).
Part of it might have been the initial enthusiasm wearing off, but I think that increasing review speed also counterintuitively decreased the time I spent on reviews each day. When I go through reviews faster, I get mentally tired faster and have to stop sooner. Additionally, going through a higher number of reviews makes me feel accomplished so there is less incentive to keep pushing myself to try to spend an insane amount of time each day doing reviews.
Lastly, here’s a graph of the (smoothed average) number of reviews per day:
As you can see, the number of reviews per day has generally gone up slightly, even though I’m spending generally less time on them, thanks to the greatly increased average review speed.
A Year of Wanikani Reviews
Since starting Wanikani (again) a year ago, I’ve done a total of 32550 reviews and spent nearly 154 hours on Wanikani. Of those, I passed 24032 reviews, for an overall average accuracy rate of 73.831%. However, that average conceals important variation.
Here is a graph of my (smoothed average) accuracy rate over time:
As you can see, my accuracy was very poor, back when I first started a year ago. It makes sense since I had tons of high level items awaiting review that I never learned properly in the first place, and I hadn’t touched WK in four months, so I barely remember most of the low level stuff anyway.
39 days in, I was forced to give up and reset from level 60 to 48, leading to an immediate jump in my accuracy rate, since I didn’t have to deal with so many of the high level items I never learned in the first place. Note that this graph is a smoothed average, and thus understates the jump in accuracy.
Soon afterwards, I reset to level 47, and then 46, which helped a bit more. However, even ignoring the resets in the first two months, my accuracy has since very slowly climbed over time. It makes sense, since when I keep missing reviews over and over, I’ll eventually start remembering them sometimes, and so I’ve gotten a lot more familiar with the items on WK over the last year. (And getting better at Japanese in general probably helps slightly as well.)
Accuracy wasn’t the only major improvement over the last year either. Here’s a graph of my (smoothed) average review speed:
As you can see, when I first started out, reviews took forever because I couldn’t remember anything. I was only managing 1.5-2 reviews per minute. Even after resetting, I was lucky to even hit 2.5 reviews per minute. Since then however, my average review speed has slowly improved, and I’m now hovering around 5 reviews per minute, and often even better than that.
These two improvements are of course closely related. The biggest thing that slows me down when doing reviews is missing reviews, especially when I check the correct answer or look up related words so I’ll hopefully remember it next time. Familiarity with the words in general helps a lot as well - having more words I know instantly and fewer that I have to agonize about helps me go through reviews more quickly.
Meanwhile, here’s a graph of my (smoothed average) review time spent over the last year:
Counterintuitively, this shows the opposite pattern. In the early months, I spent a lot more time each day doing reviews. (The big dip at the start was when I was on vacation, as are most of the other big dips).
Part of it might have been the initial enthusiasm wearing off, but I think that increasing review speed also counterintuitively decreased the time I spent on reviews each day. When I go through reviews faster, I get mentally tired faster and have to stop sooner. Additionally, going through a higher number of reviews makes me feel accomplished so there is less incentive to keep pushing myself to try to spend an insane amount of time each day doing reviews.
Lastly, here’s a graph of the (smoothed average) number of reviews per day:
As you can see, the number of reviews per day has generally gone up slightly, even though I’m spending generally less time on them, thanks to the greatly increased average review speed.