Review Calculator Idea

Wishing there was a calculator that could tell you how many reviews you need To do a day to reach a certain point, or grammar points,

EDIT*** Uhmmm Bunpro team this is a bit of traffic??! maybe this idea is worth looking into? I took a python class Freshman year so lmk if you guys want me to write the base code :sunglasses::sunglasses::sunglasses:

1 Like

You can do this pretty easily manually. For example, I’m studying for the JLPT N3. There are 217 grammar points and the test is 51 days away.

217/51=4.255

So to keep on pace, I need to do learn about 4 per day.

3 Likes

Reviews, not points. — Dave

4 Likes

One problem is that your retention probably drops quite dramatically if you add over a certain amount of new points and, equally, some points are far easier than others. This is without mentioning how factors outside the SRS change things (immersion time, health and lifestyle, etc). You can do some sort of manual calculation based on your current retention stats but it is pretty certain that your retention will drop for more advanced grammar points in general and likely even more so if you maintain the same pace as you did for lower levels. E.g., 5 new a day for N5 is fast for a total novice but it is managable, especially as you probably will be spending most of your time on studying and not just reading/listening at that level, but 5 a day for N1 is extremely quick to the point of being unmanagable, assuming an equal level of ignorance for those grammar points as you had for N5.

So…if you are going to manually calculate it then I would err on the side of caution and predict a retention drop for each level or for each extra new point you add per day. Very not scientific and janky but it would probably give you some estimation.

Just a side note but it is very easy to get sucked into the psychological trap of thinking that progress in an SRS is progress in Japanese. In one sense it is but in another sense it really isn’t. These kind of estimations are fun but just beware.

5 Likes

whoops, my bad! Didn’t read thoroughly.

2 Likes

There are review workload graphs for Wanikani. Surely someone could put their brains on bunpro too.

Though they would only be accurate assuming you got ALL your reviews correct too, right?

2 Likes

We were actually discussing at the beginning of this year about adding some sort of calculator for ‘I want to get X grammar points to X level of srs before X date’. Specifically for people that are aiming for JLPT’s or similar things.

It’s not beyond the realms of possibility. We’d just have to figure out how to take into account the amount of mistakes the average learner makes in the process of learning etc and then set up a learning path that is a ‘best estimate’ of how long something takes the typical student.

5 Likes

Could personalise it a little and use the % that particular user gets correct over time (e.g. % of items in Beginner 1 stage that user gets correct, % in Beginner 2, etc.).
Could use a Markov chain model to help :wink:
As well as adding disclaimers, like the content will get harder and your percentage will change over time, etc.

4 Likes

Is it really the case that N1 points are necessarily harder than N5? I haven’t gone through any of N5 or N4 (or N1), started with N3 and am working through N2 right now, but the vast majority of the points imo for N3/N2 are basically words that don’t have a direct translation in English and need to be explained further.

I don’t think the higher jlpt levels necessarily mean harder just more obscure / less important. I wouldn’t be surprised if earlier levels are actually harder because presumably you’re a lot newer to the language and certain things like conjugation might in practice be more difficult then some of the later structures.

Don’t mean this as a challenge btw, just read that and it made me curious if the grammar between jlpt levels are actually different in difficulty.

3 Likes

I don’t think it is slower because it is harder but that it is slower because the points tend to become things you see less and less frequently in the wild. N5 material is in literally every Japanese sentence so just by engaging with Japanese you will get comfortable with it. Conversely, some (not all) of the points in higher levels are things you might not see much depending on what sort of Japanese you interact with and, even then, it is also probably something you don’t see every day.

I actually think it is easier to pick up points after N4 since by then you know the basic shape of Japanese grammar and most stuff is just patterns or turns of phrase.

I guess it depends how you use Bunpro as well but in my original comment I was assuming that people would only add things that they are already somewhat familiar with since that is how I use it. However, having hung out on this forum for some months now I should probably drop this assumption as it seems many people add things that are totally new for themselves fairly consistently. I personally prefer only adding things I have seen multiple times and that I’m pretty sure I will have at least 90% retention for so for some points (normally formal written turns of phrase) I just don’t see them enough to feel comfortable adding them despite broadly understanding them in context when reading.

If one were doing just Bunpro and not doing much reading etc then I guess N1 might take the same effort as N5? I’m not sure as I think if you’re at that level then reading and listening are much more important than SRS and I don’t know anyone at that level who somehow doesn’t engage with Japanese as their main form of “Study”. Everyone’s experience is different though. What do you think?

2 Likes

Yeah the point of frequency of exposure outside of srs is a good one. Allows you to get extra “reviews” outside of the srs framework, and that wouldn’t be available if it’s obscure and you don’t see it in the wild.

I imagine most of the people who come to bunpro started learning Japanese for some time before, and have learned N5 and maybe even N4, which might be another reason why it has better retention rates than the others.

Your approach to adding new points is interesting but completely the opposite of mine, as I have been pretty aggressive about adding new points at a consistent schedule.

I personally feel like its most efficient to study new grammar / words before I encounter it in the wild. It’s not that I think I would actually know that point just by studying it (even if I mastered it according to SRS) since you need the context to learn the nuance and internalize it, but by studying I preload all of the difficult parts of learning the point, so that I have a good baseline to learn the nuance.

So like while I agree that reading is more important, I don’t see a reason not to consistently add points as long as it’s a manageable rate.

The only reason why I use srs more than reading is really just I can do it without thinking about how I want to study, which is important because of my other priorities. I can just dedicate some time every day to finishing my reviews and then move on. That and I’ve been struggling to find content in Japanese that I really want to consume recently, as I’m no longer as interested in anime/manga/light novels as I once was.

2 Likes

Yeah, I think you’re onto something. Having some awareness of a grammar pattern beforehand definitely makes it easier to spot in the wild. I used to do a similar thing with vocab although now I generally mine. Main thing is to be doing anything that keeps you studying and motivated though. It is fun to have discussions about efficiency etc but, honestly, the most “efficient” thing is just whatever keeps you going.

2 Likes

For me, I think the other way around would be most useful - if I want to average x reviews per day, when should I study new grammar points and how many? It’s very easy to end up overwhelmed with reviews if you go too hard on adding new ones. Obviously, the answer will depend on how many you get right vs wrong, but I think you could get close enough with some fudging.

2 Likes

yea that’s actually a good point. I could see both ways being very helpful depending on the situation. THIS COMMENT TOO BOYS ^^^^^^^^^

Yeah this sounds very useful. A friend of mine tried out srs, but only wanted to introduce new cards that would keep daily reviews at a low number. I couldn’t figure out how to make an estimate with that app, but for people like him it would be immensely useful.

One thing that might add difficulty to making a calculator is how often are people doing their reviews. Like we know the interval for beginner 1 is 4 hours, but how often are people doing their reviews after 4 hours (as opposed to say once a day). Do we make the calculator with the assumption people do their reviews at the optimal interval schedule?

I think at least for the early levels, the amount of reviews you do in a day varies wildly with how frequently you’re doing reviews, although I’m not sure to what extent this is the case or how it would affect a calculator.

There are 12 SRS stages on bunpro, so if you learn the same number of cards every day you’ll eventually get to at least 12x that number of card as daily reviews, assuming no mistakes.

In practice of course there will be mistakes that will increase the amount.