Is accuracy this low normal?

Not the original poster, so this is just my interpretation.

My thinking is the meaning here is ‘overlearning’ beyond the expectations of the SRS interval system. The system is expecting you to only see a specific query only at specific intervals and is optimized to be effective in that vaccuum. Exposure outside of that is ‘more reinforcement than the system expects’, or ‘overlearning’.

I think thats a different connotation to ‘overlearning japanese’. I think learn as much Japanese as you like and more reinforcment can only be better. However, you’ll produce a higher than the expected accuracy than SRS alone.

So i guess ‘overlearning’ is in reference to the evaluation of efficacy of a learning system which expects specific input.

‘Study’ is a general term for time spent learning, or the action of learning. Again, I don’t think the user would ever suggest to limit your study.

Hope that helps!


Thoughts on “overlearning” - the reason it’s called that is opportunity cost. High accuracy indeed suggests that you could widen the SRS intervals to be able to learn more material in the same amount of time.

True, however consider that the time you spend reinforcing could also be time spent on learning new material, etc. Balancing efficiency and enjoyment is a tricky part of any kind of study, so I’m not going to claim that going at max efficiency is necessarily the best plan, but I think it’s a good thing to be aware of.


well said!!

I am studying everyday outside of WK, practice speaking every week, grammar on Bunpro and working with some italki instructors… and of course reading is fun (when it’s not a horrible wall I crash into)… the only reason I haven’t quit WK yet is because I’m still learning…

If you are feeling frustrated, burned out, etc…do something else…put go slower on WK, pick up some manga, book clubs, whatever…find something to keep your interest…

Yeah totally agree!

One thing that’s interesting to think about as well is ‘distress tolerance’. Different people have different levels of distress tolerance for failure states and it tends to sit on a bell curve. While ‘80%’ might be optimal in terms of optimal memorisation, it might be too high of a failure rate to keep a individual motivated. Whereas another individual might need a 30% failure rate to feel appropriately challenged.

Interestingly, people tend to overstate their ‘distress tolerance’, meaning we often need more positive reinforcment to stay motivated than we realise.

1 Like

I know that has customizable SRS intervals and is generally a good website

tbh i sorta wish I had discovered it instead of WK because i’m only capable of doing one SRS system at a time and it’s just better lmao

too late to stop now tho lol

There’s a certain accuracy percentage that you should aim to exceed with your reviews, but I can’t remember exactly what it was. Something like 66% or 70% accuracy. If you get below this point regularly, you won’t clear out the items faster than they’re coming back, and it will be hard to progress. If your accuracy is higher than this, you should still be making progress, though you’ll have to put a lot more time and effort into WK than someone with higher accuracy, because you’ll have to do a lot more reviews overall in order to burn the items.

The big trick to increasing your accuracy, I think, is to spend more time on your lessons, and to hit the first two review stages after that very close to their SRS intervals. Some people spend a lot of time looking at the mnemonics when they learn the items and make sure they’re pretty well cemented in their memory (and review their mnemonics when they miss a review). Personally, I drill the new lesson items immediately after learning them by running through the self-study quiz script. After I started doing this, my accuracy increased.

The later stages are a little more flexible than the first few apprentice stages. If you wait a day or two extra on an enlightened stage item, it’s not really going to make much of a difference, but if you wait a day or two extra on a first stage apprentice item (due in four hours), that’s many times the review interval, so you will struggle a lot more to remember that item. Hitting that 4 hour review interval and then that 8 hour one right after will do a lot to increase your overall accuracy on those items.

And yes, exposure to Japanese in native materials or in other forms of study outside WK will also help you remember the items and increase your accuracy, but in my opinion, the two most important things to do if you want to improve your overall accuracy while putting less time into studying are 1) making sure you learn the items well during the initial lesson stage, and 2) hitting the 4 hour review and the 8 hour review intervals as close to those times as possible after doing your lessons.

1 Like

I don’t really think accuracy % matters anyway. As long as you learn the vocab in end, you still reached the goal. Don’t worry if it takes you 1000 tries to get it right.


I see, thank you!

It’s a pretty neat site!

My only caution for those with low accuracy: beware of your number of daily lesson, as the burden of downgraded items may run over you and induce a burn out. Keep it slow!


Oh that explains why I can maintain my 95% accuracy, though, I wouldn’t call it “overlearning”. I just spend more time in the language that I fell in love for.


Exposition is absolutely key.

人間 is one my favorite examples, it seems so weird at first, but any fantasy anime or game will absolutely spam you with the word.

Another one is 現在, I never Really remembered it means “present”. Well put your decides to Japanese and you’ll see 現在オンライン all the time.

On another note, Google Maps in Japanese is Katakana madness, so good practice too.

1 Like

I don’t think it’s overlearning, either. I think what was meant with overlearning in this context, and what it usually refers to, is spending extra time with the SRS items; doing leech training, spending a lot of time with the specific items only etc. You are “overlearning” the material by more than the system expects and what should be necessary to learn them. That time could potentially be spent better for other areas of Japanese like immersion, grammar etc. That’s why it has a slight negative nuance.

It’s the same concept when people recommend to have your SRS accuracy around 80-90%, because otherwise efficiency drops rapidly. On the other hand, my mature card retention in Anki is only slightly above 70%, but it allows me to overall learn more words when I have room to add them more frequently without worrying about accuracy.

@shuly Feel you on the leeches. One of the main reasons I moved to Anki after reaching 60 for a while. I feel like the leech problem becomes worse the slower you go. I did WK in 1,5 years, so the leeches didn’t have time to accumulate as much, but someone going a lot slower will create quite a wall of leeches that can become insurmountable. Hard to add new lessons at that point.


hmm… that’s interesting… when you say slower? do you mean slower between review cycles or slower as in lessons?

Overall I’m not missing the review windows for the App1-Guru1 / the G2+ vary a bit because of the timing… what I’ve found is that I forget them (like I’ve never seen them before when they hit G2/Master/Enl)…only way to manage that would be with reviews SOONER before you forget them (the whole point of SRS).

I’ve slowed down so that the leeches don’t pile up and take over 80+ percent of the reviews and this is unfortunately the pacing that I have settled into for now… why won’t WK build a freaking leech manager ugh!.. anyway… my pacing now is 40-45 days per level doing lessons and targeting next day reviews about 100 for new lessons…(limiting new lessons to 7-8 a day max) when I can do lessons.

after a long slow slog and trying all the scripts / tests / etc…nothing really worked so after doing what I could got enough data to average it out over time so that’s what I’ve got.

What would be helpful is more people with low accuracy and/or have to go slow to share their stats/experience … it would be more motivating to hear from those (recently there was someone yesterday 6 years)…that took 4-6 years to get to level 60. We all can’t go fast and not all of us are good at languages…I’ll take math, chemistry, and engineering any day of the year over languages!

1 Like

Sooo much this! My drop is not as dramatic as yours but clearly visible after Guru 1…

That’s why I’m really grateful for Anki where I can tune my SRS timings to give me accuracy I want (and not the other way around…).

I had my 4-year-anniversary last month, and I’m at level 52 now as you can see. :relaxed:


To the OP, I’m going to disagree with some of what others have said. 75% is pretty low. While you shouldn’t stress out by comparing your recall rates to others, I think this is an indication that you’re not forming a strong enough mnemonic memory association.

For example I’m usually around 90% recall rate, but if I find myself dropping to 80% or lower then I spend some extra time visualizing each of my failed kanji at the end of the session, to solidify their stories. It almost always works to bring my recall rate back up the next time around.


I just meant in general, as in the slower you take to reach lvl 60 and “complete” WK. Maybe it’s just that the later items don’t have enough time to become leeches and bog down the apprentice count. Another factor could be that the more one increases their immersion and reading time, the more leeches just tend to disappear.

In Anki I have a pretty strict leech policy, where I suspend them automatically after 5 fails (in review state, not learning). Unsuspending a year later, they often don’t bring any problems anymore, since I learned them elsewhere. So I usually don’t even bother with them, because there’s still so many more words to learn instead, which is more efficient.


see when you say anki, do you mean the free ankiapp or the one that’s like twenty dollars?

so are you telling me to study the kanji and vocabulary after my reviews?

Just Anki It’s free on desktop, but I have bought the ios app, as well. Well worth the money.

AnkiApp is a knockoff riding on the branding of Anki. Doesn’t have nearly the feature set of Anki and is honestly just doing shady practice. Wouldn’t touch it with a ten feet pole.


okay, thanks!

1 Like