Anki Intervals Between Reviews (Vocab)

Anyone have a recommended interval between each review for vocab? As of right now I just copied WK’s intervals (4hrs 8hrs 23hrs 47hrs… 7 days… etc). But, I was wondering if anyone recommends different intervals. (I’m doing N5 sentence vocab cards if this makes a difference.)

Thanks!

Well, for my non-WK deck, I have the steps for new cards at: 1, 60, 180, 480 (these are minutes). Once it hits 480, then it graduates from “new” to “review”, I have the graduating interval at 2 days, and I have the ease set at 250%. This means the intervals are usually: 1 hour, 3 hours, 8 hours, 2 days, 5 days, 12 days, 30 days, 78 days, 195 days, etc. and it keeps going by a multiple of 2.5x. Of course the intervals aren’t always exact since Anki does sometimes move them around to try to even out your workload (which sometimes works counter-intuitively, but oh well). Apart from the “new card” intervals and the graduating interval, the ease itself is Anki’s default.

1 Like

I read of a SRS system once that automatically adjusted the times based on your success rate. Both globally, if a certain interval seemed to be a little long or short for you, and also on an individual-item basis because some are harder than others.

It turns out though, that it doesn’t matter as much as you think to fine-tune that. It just matters that whatever you pick, you DO it.

2 Likes

That makes sense, I haven’t missed any reviews or lessons since I started about a month ago in RRTK. I’ve had pretty decent retention rate of 1000 kanji (about 95-96%), I just didn’t know if it would be different for vocab.

i believe wk’s lessons aren’t thorough enough. i wouldn’t rate “got it right once” “learned”.
i’d rework this initial step into a little drill that only releases the item after 5 or more successful reps, because doing a lesson, getting it wrong, then being tested on it again once immediately isn’t effective for me. i need to grind some things more to make them stick.

i’d also shorten the review intervals some, implement a few more steps, and abolish “burned”.

1 Like

I’d be good with that, except for the abolishing burned. It’s a system, there’s an inflow (lessons) and an outflow (burning). If there no goes-outs, your workload is going to get out of control.

Maybe that’s the solution, add a stage that waits, say, one year. The strategic deterrent interval. You better finish, or else you’re going through everything a second time. :slight_smile:

well yeah, the intervals would keep growing to some degree. while items will never go away completely, you’d only see them once in a blue moon, at some point.

Well you would see each item once in a blue moon, but you would start to see items (plural) at the same rate you got them into that state. Which is the same rate they got to the previous state, and the previous… long story short, it’s the rate you do lessons. It would be another simultaneous repetition of everything. There are already 8 of those.

So I guess you could say well then, a 1/8 increase in work isn’t that much. Or you could say, uh, 8 x the lesson rate is already plenty, thanks. I’m in the second group.

The thing about burned items though is that you’re most likely going to see that in the future, outside of WK or Anki or whatever else you’re using. So like there’s really no point in getting a card in 6 months because within that time period you’re almost 100% guaranteed to see it in your studies.

I don’t use WK at the moment since I went a different learning method, but when I was learning the 1000 most common Kanji I used WK’s intervals and I would learn 30-40 new kanji a day. It typically took me about an hour or so to learn the kanji, and I would set an interval for 10 minutes and then 2 hours. After I learned all the new kanji for the day and did the 10 minute interval review, I set my “learn ahead” to 2 hours that way I would see all the new kanji I learned that day at once, without there being too long of an interval in between. I would get mostly 28/30 correct. Those that I got correct I did a quick review of how I remembered it and would go on to the next one. The ones that I got wrong I would figure out how I tried to remember it, and then created a story based on the way I was trying to remember it. This was a great learning method for me and I’m maintaining 96% retention rate of 1000 Kanji that I learned in 23 days.

i believe that timers here need to be a bit more random anyway. you can schedule items earlier with no ill effects for their state as memory.