Stop being "punished" for not doing a review on time


#1

Currently for an item to progress from apprentice -> burned you must correctly answer the meaning+reading of it 8 consecutive times at varying minimum time intervals.

The intervals are: 4h, 8h, 24h, 3d (guru), 1w, 2w (master), 1m (enlightened), 4m (burned)

This system works OK if you are able to consistently do your reviews but isn’t as efficient and flexible to people who aren’t able to complete their reviews on a regular basis. I just wanted to know if it’s possible + if you think it’s a good idea if the review items were to account the last time an item has been reviewed in order to place it in the correct level of comprehension based on the time between reviews.

An extreme example to showcase this is, say you unlocked the vocab “一” and have memorized its meaning and reading. If you take a break never doing a single review of ”一” and come back and review it after learning it exactly 5 months 3 weeks 4 days and 12 hours ago (the minimum amount of time to get and item burned) and get it right, with the current system it will only go to Apprentice 2 but what I’m suggesting is that I think it should go straight to burned as you’ve clearly shown you’ve memorized the meaning of the word and require no aid from the SRS system.

A more realistic example that showcases the benefit of this system would be, if you unlocked the vocab “二” at 9am and you were really busy all day and had no time for reviews until you got home and you then reviewed it for the first time at 9pm, with the current system if you got it right, it would also only go to Apprentice 2 and the next review will be held 8h later, but if you take into account the last time this vocab was seen 12 hours ago (4h + 8h) I think it should go straight to Apprentice 3 and the next review will be held 24h later.


#2

In practice, your suggestion would only affect the Apprentice levels, and I think there’s a likely problem with that: You may still remember an Apprentice item after a few days or a week without reviews, but your long-term recall will be worse. The frequent exposure during Apprentice reviews is the most important part of the memorization process.

Also, the ‘punishment’ for delaying reviews by a few hours, or half a day, isn’t really that bad. If you review only once per day, you’d still level up in 8 days rather than 7. That’s only 60 extra days to complete all 60 levels.


#3

I’m in agreement with rfindley here. And what if you really only got it by luck? Without having the extra reviews to verify that, you might go around thinking you were right with your reading, perhaps, but there is or isn’t an extra う (this gets me a lot of the time… remembering which do or don’t have longer vowel sounds… most recently with 登山 because the kanji alone for 登 can be read as とう or と). It’s little things like that that later on can make a big difference.


#4

I agree that this will mostly effect apprentice levels, but I don’t understand how it will make long term recall worse as it will still follow WK’s SRS intervals when you get the item wrong/right, it just reduces the frequency that an item is tested if you have shown that you have remembered it without those reviews.

That’s true, but if you are able to show that you can still remember the meaning of the word after some period of time is the frequent exposure still required for the memorization process? Because if you get it wrong the item would still be treated the same if you get an item wrong with the current system.

For me personally, I’m getting to the point where my grammar/vocab knowledge is enough that I can go through basic manga and spending more of my “study time” on consuming native material than doing reviews, I don’t think the frequent exposure in WK is necessary to help memorize vocabulary.

Punishment was used for a lack of a better word, maybe penalty is more suitable. Also how did you get 8 days for level up time if you’re doing 1 review session per day?

From my math it should be: 13 days
Day 1: Item lessons
Day 2: Apprentice 2
Day 3: Apprentice 3
Day 4: Apprentice 4
Day 7: Guru radicals/kanji + lessons for second batch of kanji
Day 8: Apprentice 2
Day 9: Apprentice 3
Day 10: Apprentice 4
Day 13: Level up

With my proposed solution it’s: 11 days
Day 1: Item lessons
Day 2: Apprentice 3
Day 3: Apprentice 4
Day 6: Guru + lessons for second batch of kanji
Day 7: Apprentice 3
Day 8: Apprentice 4
Day 11: Level up

So it’s a 120 day difference not 60.

How can you even get the reading for a word right by luck with the current system? Because WK right now requires that you type in the exact reading so there’s no chance that you can even input an extra う and have the result be correct.
This system only cares about the items that you get right after a long period of time without reviews, if you get it wrong it’ll just be treated as any other item that you get wrong.


#5

If you don’t really remember it and you sort of guess. That’s what I mean by luck.


#6

But you can’t get the reading of an item right “by luck” as it requires typing in the exact reading of the word, so the example you provided in your previous post:

has 0% chance of happening as WK requires the exact reading of a word to be inputted, you can only get the meaning of an item right “by luck” so I don’t understand how what I’m proposing worsens a person’s ability to recall an item.


#7

She means you have the option of, for example, putting とうor と. You don’t know which it is, but you do know it’s one of them. You make a 50/50 guess and get it right. That’s luck.


#8

I know, I understand what she is saying but in her example for 登山 if you were to input とうざん WK would mark it incorrect and only accept とざん so if your purpose is to learn kanji+vocab why would you guess the answer in some hope to get it right when it is obvious you don’t know it and should just mark yourself wrong in the first place you’ll just be lying to yourself when you guess and get it right.


#9

Are you actually saying you’ve never not remembered the reading of a kanji, guessed こう or ちゃう or one of the other common readings and been right? Because if that’s true then you are probably in the minority.


#10

I haven’t been using Wanikani since last year. And back then the fastest level up was almost 8 days. (7d 20h if I’m not mistaken) You are talking about leveling up in 7 days. Is it 6d 20h now? And if it is, what’s shortened?

Sorry,
Edit: @rfindley You were the one who said 7 days. Something changed?


#11

Sorry, I was also thinking of doing Lessons and the first 4-hr interval on the same day, though I realize that’s not what I said.

Two mistakes in your calculations:

  • The 4th Apprentice interval is 2 days, not 3. That’s now true for both Radicals and Kanji [edit: and vocab].
  • You have to start on Day 0, not Day 1, because the first day on a new level is also the last day of the prior level.

But either way, the repetition early on is important in how neural circuits work. Synaptic strength is reinforced through repetition. Suppose you see an item during the Lesson, then come back 3.5 days later for the Guru review (using your proposal). Even if you remember after a few days, it is a weak connection. You may not be able to sense how weak it is, because the synapse merely has to fire for you to remember. But since it is weak, you are closer to forgetting. But if you are exposed multiple times during those critical first few days, the synaptic strength is much higher, and you will remember longer.

Frankly, SRS is a tradeoff: It saves you time by minimizing your required exposure, but it comes at the cost of strength of memorization.

[Edit: By the way, I’m always in favor of trying out new ideas, and this one is no different. The above is what I suspect the result will be, and I think that’s sort of what you were asking for?]


#12

@GCA, Yes, they reduced the 4th interval for kanji from 3 days to 2.


#13

Well I dont really see much benefit to guessing by typing in one of the common readings so Ive never guessed the reading of a kanji that I couldn’t remember, I’ve only ever guessed if I’ve confused which reading is the onyomi and kunyomi.


#14

I find your comment slightly bizarre, to be honest. Of course I guess. I guess all the time. That’s how I get stuff wrong. Are you suggesting you only answer if you are completely sure you’re correct and put in something deliberately wrong otherwise? Because my review scores would be sub-50% if I did that.

Plus, if you knew what she meant, why did you spend time explaining how: [quote=“DiMono, post:6, topic:16699”]
you can’t get the reading of an item right “by luck”
[/quote]

and it:


#15

Great news! Thanks.


#16

Didn’t they reduce it to 2 days for vocab as well? That’s what Kristen said at least. I haven’t actually bothered to check.


#17

Yeah… I didn’t mention vocab in this case since it doesn’t affect level-up, but you’re right.


#18

Then it’s about time they update the srs graphic in the guide. It’s been almost 3 months now.


#19

Ah my calculations were from the intervals shown in the WK guide, completely forgot they changed it recently.

I see what you’re saying on the value of repetition, but is the same amount of repetition required for each item? I don’t know about you but some items are easier for me to learn than others. Going from your example if I did the lesson and came back 3.5 days later for guru, if I had a supposed poor synaptic strength then I should fail the review (where the chances of failing would increase the higher the interval rate) right? Meaning that WK’s SRS will kick in and correctly reduce the SRS level thus increasing the frequency.

Please let me know if I misunderstood what you said.


#20

Yes, I deliberately enter something wrong if I’m not completely sure about the answer or if I take too long to think of the reading+meaning since I like to pretend a native speaker is right next to me asking me the meaning and reading of each item. It hasn’t really tanked my review scores, according to idigitech my overall score for kanji is 91% and 90% for vocab