It is been a while I am using Wanikani, and I really like the recent changes regarding lessons and especially the fact we can control the number of lessons we want to do per day.
I had to restart Wanikani once or twice because at some point I accumulated too much lessons and preferred to reset to some previous level. To prevent that, a strategy I tried is to not go over some limit of accumulated reviews for the next week that sounds reasonable to me.
Now that we can fix a target for the number of new lessons per day I would like to configure this number to adapt to my needs so that I donât get overwhelmed with reviews.
However, I think it would help me if it was possible to get an approximation of the time it would take me to do the reviews instead of the number of reviews.
For example, along side the number of reviews per day, an approximation of the time taken could be printed. This time could be computed following a simple formula (average time per radical * number of radical + average time per kanji * number of kanji + average time per vocabulary item * number of vocabulary item).
I guess, once we have that, it is also relatively easy to have an estimation of the time taken for a given number of the new lessons per day if we take into account the success ratio rate of the review.
This way, we could say that with 15 new lessons per day and a ratio success rate of 100% you need to spend in average 1h on reviews per day (± 1h). This would really help to set this number correctly. I suspect for this number it is not that simple, but using the model above and doing some simulation, we can have an approximation.
Finally, I wonder if at the end, more than setting the number of new lessons per day, we could set a target of the time we want to spend of reviews, so that the number of new lessons per day change to adapt with the time we want to spend on reviews?
Yeah itâs really not because people vary widely with how fast they do reviews and even each person varies day-to-day. Youâre doing the best thing you can do to even out the variability by doing a set number of lessons every day. But still itâs going to be plus or minus quite a bit.
That said, once youâre in âthe zoneâ, consistently doing lessons at some rate and reviews in all the SRS stages coming in, a useful thumbrule is however many lessons you do a day, youâll have 8-9 times that number of reviews. So I guess you could guess how many reviews you will have, and estimate for yourself how much time it would take you to do that many reviews.
This is even more complicated when you change something. If your failure rate changes or you start doing more or fewer lessons per day, your review load will also change - but not instantly. Thereâs a lag time, and changes you made up to 4 months ago might be affecting todayâs number of reviews.
** ON AVERAGE **
Again, if youâre not remarkably consistent in the lesson rate, that average isnât going to be a very useful number, because the plus/minus will dwarf it. If you donât have the capacity for the âplusâ part of the âplus/minusâ, thatâs still a problem, even if the average is achievable.
I would be curious to know the variance, but I think having an estimation would be rather nice. If the estimation is an over estimation or an under estimation it might be quite easy to adjust in practice per person.
If you donât have time for the plus, you can always prioritise reviews so that you donât do new lessons. In a few days, you should catch up relatively quickly.
The suggestion is not about the platform committing on a time, but rather an indication to help with not being overwhelmed of reviews.
I would be curious to know the variance in practice with the very simple model I proposed above.
I think youâre missing something important - that what matters for not being overwhelmed are not the reviews or the reviews time but the lessons. Lessons take the longest portion of your time and thatâs what you should measure - how long does it takes you to learn new items not how long it takes you to recall them. The more time and attention you put into your lessons the better your success rate will be.
The question is what is the number of lessons you can afford to do time wise, and whatâs the point where the amount of new items is to much for you so your success rate goes below 80%.
Keeping you apprentice below a certain number is pretty straight forward, you do x amount of lesson and regularly or stop doing them if you go beyond your upper limit target number.
Doing lessons is a different process and paying attention to whatâs your capacity to deal with new material will help you keep your schedule more in checked, so when you canât deal with new lessons, youâd be able to still do your review with a high success rate that doesnât overwhelm you.
Maybe because I am using different tools for learning Japanese, personally, I donât spend a lot of times on lessons. I would be curious to know how much time is spent in average by each user while learning a new item.
However, I do not buy your argument that it is the lesson time that matters. When starting a new lesson, it is relatively easy to know how much time you will spend on it. It is less easy to know the time you will spend on the reviews and when. This is even more less obvious when you take into accounts the reviews for other items.
Even though you can make approximation like 20 seconds per vocabulary item * 10 to reach the burned state, it does not help you to know whether you will be overwhelmed or not, especially because the reviews are spread out over months.
However, I have the feeling this could be approximated (and hardly by a human but with an algorithm).
I try to keep my review time to a bit over an hour a day.
Sometimes that is because I did all 15 lessons for the last 3 days in a row or sometimes that is because Iâm struggling with my current crop of reviews. Either way I found that I start to get frustrated with the review grind if it takes me more that 2 hours for multiple days in a row.
Iâm sure this differs wildly per person but Iâve found that is the sweet spot for me that still leaves time and mental capacity for doing some listening or reading practice.
The formula for how long it will take you to do your reviews is a very simple one since the intervals are fixed and itâs just a matter of putting in an estimation of percentage failed reviews which increases the amount of reviews and gaging your personal reviewing time per item.
But again, this formula does not take into account how long it takes you to learn an item.
Iâm writing this from experience, as someone who did a reset a couple of times and changed the way I do the lessons - lessons are the most important part to put your time in and to do it slowly and with a lot of attention, the less attention you put into lessons the longer your reviews will take.
20 seconds sounds like a lot for an item, ideally you should be able to read a prompt and type your answer immediately, which shouldnât take more than a few seconds. What determines how long the recall process takes is how elaborate was your memorization process which happens in the lessons part.
You can probably poll the forum to make a small survey of how much time people dedicate to the lesson part, but at the end of the day you should be able to see for yourself what difference the amount of time and the amount of item per lessons do to your personal accuracy level. Something like, 3 is too easy but 20 is too much.
Statistics is nice but it canât tell you where you fall within it, thatâs something that has to do with observing your tendencies.
Recently Iâve made myself think of my own example sentences when I do lessons, using the vocab and sometimes grammar I recently learned. So, a tiny batch of 3 lessons takes me 5-10 minutes. But on my reviews I answer them without thinking too much. So not very many seconds. Sometimes if itâs on the tip on my tongue (fingers?) I go make a cup of tea and answer it when I get back. That takes longer and would skew the average quite a bit.
You are making this too complicated. Firstly, measure how fast you can review items on average, as that varies greatly person to person. You can do this by timing how long you take to finish a batch of reviews. Try to use batches with at least 20 reviews for this. Measure several times and after a while you can get a pretty good idea of your typical pace. For instance, my pace for reviews is about 3 to 4 items per minute, when on my laptop (much slower on my phone). If there are 30 to 40 items in the review pile, I know it will take me about 10 minutes. If there about 100 items, I know it will take me about half an hour.
Then use the review forecast to estimate how many reviews you will be getting and whether you need to decrease/increase the number of lessons for the day.
This is the best way. Time 100 reviews, donât try to time one and extrapolate that to 100. Make sure the batch includes a representative sample of items in each SRS level, and some previously-failed ones.
Thatâs what I wouldnât do. I would let the +/- variation just be and power through the reviews. If I felt that it was consistently too much, I would make a change, but not be too quick to react to dips and spikes. Every change you make induces more fluctuation and wider swings.
(This is also why I like a âlessons per dayâ rule, not a âapprentice item countâ rule. The start/stop nature of the latter trashes your consistent flow.)
Matter of personal opinion, obviously. Canât really say anyone is doing it âwrongâ.
Itâs a wanikani tradition. âShut up and just do what it tells youâ is a valid strategy too. Arguably, the best one, since weâre all japanese learners not learning system professionals. But whereâs the fun in that?
Yeah thatâs a HUGE source of variability - some people try to get the correct answer on the review even if it takes some sitting there thinking it through, trying to recall the mnemonnic, etc. Some people just put âăâ and force a wrong answer if they donât immediately remember. Both are âcorrectâ ways to use the system.
WaniKani will of course allow us to spend as much time on reviews as we put into them â they will keep coming back as fast as you submit them, and as a result can pile up beyond our ability to actually learn the items if we donât take a measured approach.
Thatâs why I prefer setting a sustainable goal to stick to (on average) daily, so I donât stress myself out by accumulating more reviews than I can keep up with, as that ultimately makes it harder to progress, due to forgetting more of the items further before being able to get to them again in my review pile.
And as far as an amount of time per item, personally I find that it varies wildly, though Iâm sure what takes me longer may be quicker for others and vice versa. Cheers!
