@sweetbeems I’m surprised you haven’t updated your avatar here on WaniKani to be the Natively favicon or something like that.
Agreed, I should have something. No I do not currently.
Wow! I do really like this and all your suggestions for the K value! I’ll think on it. I am just trying to weigh the value of doing a switch like this with potential complications vs just continuing on with my V1 initiatives (community / audiovisual). I think I do lean towards coming back to grading updates after my other V1 initiatives just because I do think the current grading system with the 6 grading limit is pretty ok and very easy to monitor, but it obviously something I’m continuing to think about. I know the grading limit is annoying and especially not being able to grade the things you want.
Heh, I’m not sure if anyone actually reached 100 gradings on one book before the gradings limit, even Naphthalene. However, a few people reached 20 / 30 which already gives you a tremendous amount of power (can alter a book 4-5 levels potentially). 100 was just a round number on my part , but you are right that the adjusting levels does open up more gradings!
Edit:
Very well… shamed into adding an avatar
I’m not saying you should update the algorithm as the very next thing you do. But I definitely think you should flesh out the algorithm before moving into other mediums. I think you should get the Japanese book use case handled as well as possible before expanding.
You’re probably right that I should at least figure out something which allows you to grade a bit more. However, I think improving the accuracy probably is less important as I think it’s already relatively good.
The two very well could be correlated.
Yeah, I know we’ve been talking about this a lot the last handful of days. I do agree that the current approach isn’t bad, for now at least. A large part is that I do want to do more comparisons, regardless of the details of how much that impacts the levels overall. I’m sure a lot of users feel the same.
In my ideal world, you’d force (barring skips) a certain number of comparisons for a given book/series, such as the 6 you already have. But you’d also always allow the user to request additional random comparisons for a given book, plus allow them to manually select any pair for comparison. Again, this is my preference because doing the comparisons is a novel approach (to me) for comparing difficulties, so it’s interesting to think about. And of course I hope that additional comparisons would (given the right algorithm) improve the application overall.
I hope I don’t come off as pushy. I just like to talk through the problems and give suggestions. I get that it’s your application and you will ultimately decide how it works.
No not at all! It’s why I do spend a lot of time engaging in this forum and (hopefully) I am being very transparent. This sort of feedback is very important and helps me gauge what to prioritize.
And I agree! Your perspective on what you find valuable with Natively (gradings) is important, especially since you’re contributing so much to the platform! I do want to support you and keep you excited about contributing.
I just also have to be a bit wary that we don’t focus too much on a few things that aren’t working optimally but ok vs big changes that’d bring Natively to a lot more people (like community / audiovisual).
And to be clear, this is all about prioritization which hopefully won’t make too much difference in when things come out, only a few months. But I am hearing that this is probably a bigger want (more gradings) than I initially thought, so I do think I’ll do it sooner🙃
まちカドまぞく’s level is no longer temporary, making it officially the hardest manga on the site. (or
depending on your view)
I gave up on reading it after a few pages.
Looking through the search page with novels included and sorted by level descending made me realize just how many high level novels have temporary levels. I noticed there are some additional filters like “Free” for the search page, so maybe it would be nice to have a filter like “Exclude temporary graded books” or something like that.
And by the way, ドグラ・マグラ has a temporary level of 51, which is outside of the allowed range.
That’s a good idea on the filter, would be easy.
As for the level 51, I never had particularly strong opinions on limiting the highest level to 50, so if things go above that’s alright with me. Granted I do need to add a difficulty level filter for 50+.
I am curious if there really will end up being many novels that high. 雪国 seems like the hardest book a lot of people have read, even very prolific readers, so I would be surprised if there were.
Yeah, I was just going by the fact that most places on the site can only filter up to 50.
I also don’t like using the term “Native” for any of the levels. It implies that it’s unattainable, depending on how you think of “native”.
I agree with this - I also am not a fan of using ‘native’ for the same reason, but acknowledge that there will still be challenging content even for someone who passed N1 (based on talking to people who’ve taken it - I never have and probably land at ~N2 still ). I wonder if N1+ might be a clearer name?
I believe I use ‘fluent’ for 50+, which I agree is another nebulous term but I felt wasn’t unreasonable if you’re reading stuff harder than 雪国 and seems attainable. No strong opinions though, could simply do ‘very advanced’ or simply ‘advanced’ and drop the ‘fluent’ if people prefer?
If I do use the term ‘native’ anywhere, if you could point me to the spot?
Edit: N1+ also works
I just checked the 2 places I remembered it being and you’re right, it says ‘fluent’. I do remember not liking the term for post-N1 though, and possibly just conflated it with the name of the site. For myself ‘fluent’ has a similar “unreachable” feel to it.
Yeah, I must have been thinking of fluent. Oops. I’m still not a fan of the term though, and would rather it just be another Advanced level range.
Sorry for breaking some expectations here, but I’d like to clarify this a bit. A short while ago I listed the aspects of “difficulty”, and Snow Country is a prime example that demonstrates how important it is to know which of those aspects a grading is mainly based on. While Snow Country was written in the 30’s, it is still written in modern Japanese in a pretty accessible style. Old-fashioned grammar or kanji occur rarely (if at all), for example. There is no need for N1+ Japanese skills to understand the language being used - I think N2 would be totally fine. Of course, a bunch of vocab words may be unknown because they are old-fashioned or rarely used in other books, but that’s what at least I expect for basically every book I read, so that did not bother me as much. So, why is this book being graded as the hardest book on this platform? I’d say it’s mainly because it’s literature. The writing style is very vague and a lot is being left implicit. There are jumps in time within a passage of narration and between passages, people refer to events without clearly addressing them, sometimes it is not even clear at first glance who is speaking, and to whom. This leaves a constant fuzzy feeling of “what on earth just happened?”
I found a Reddit post which addresses this issue as well. Money quote: “I don’t know if it’s just the way that Japanese translated into English works or if it’s just this writers style, but everything feels very fragmented and confusing. I’m like, a tenth of the way in and I’m not even sure if there’s one plot-central female character that the main guy is interested in, or two.” And this person is reading the English translation!
TL;DR: There will be many, many books out there that are much more difficult than Snow Country when it comes to the language level…
Very interesting - I was wondering why people thought it was hard! It does seem that for these more difficult novels the content/specific writing style seems to be a large driver of difficulty and (possibly) could be a large source of variance. One number certainly doesn’t specify why something is difficult that’s for sure!
One could argue, I suppose, that the JLPT style levels is counterproductive. While it may be beneficial to first-time readers, I suspect most people will simply check the levels for things they’ve read before and find something at an appropriate difficulty relative to that. So in that sense, maybe just having the level numbers is sufficient.
Yes, I was thinking along the same lines. I think for the beginner levels (i.e. around N4) this might be of some value, although books written for natives of course don’t strictly adhere to those levels. But everything above this is somewhat doubtful at best, misleading at worst. There are simply no N3 or N2 or N1 books (that are aimed at natives) in the wild
I do agree that the approximate mappings are… very approximate. I do wish I could do some extra large ‘~’ before and perhaps it is counterproductive and I could ditch it entirely.
However, for new users to the site I think it’s a little convenient to have some very vague sense of what the Natively levels map to. More experienced users will just be comparing the levels vs things they know and generally disregard the JLPT level (at least I think so?).
My perspective is I’d probably only ditch it if people found it quite unproductive, but I have found that newer users like the JLPT levels and find them somewhat useful
I also won’t lie - it’s quite convenient to be able to list a link to ‘n2 books’ on other social media platforms. Yeah, it’s approximate… but I think it’s just as approximate as what other people are doing?
It’s just hard when nobody knows your difficulty scale.
Hmmm, so… what would that actually mean? First of all, JLPT tests for various skills. The most fitting ones in the context of books will be kanji, vocab and grammar, I guess?
For vocab, I think we can easily agree that no book (that’s aimed at natives) will strictly adhere to the vocab generally associated with a given JLPT level. I tag my book-mined vocab with its JLPT level, and for each book I’ve read so far, my mined vocab ranges across all JLPT levels, and lots of words are not even associated with any JLPT level. So we can simply disregard vocab, I guess.
For kanji, if you look at books aimed at young children, the number of kanji will usually be quite restricted, but that usually makes these books harder to read :-/ From a certain age group (maybe age 14 or so??) onwards, kanji usage will be unrestricted (usually non-joyo kanji will get furigana, though). For adults‘ books as well, of course. Now, the joyo kanji are somewhat equivalent to N1, I think? That would mean basically all of those books would fall into the N1 bracket. Which clearly doesn’t match their current Natively level (see e.g. Yougisha).
Finally, we have grammar. You may find N1 grammar points even in books for small children because they of course know a lot of grammar just from learning to talk. N3 might be a good basis for starting to read normal books, but the more N2 and N1 grammar points I learn, the more I discover them in books… So there is no clear cut between grammar levels either.
So, what do you think a list of „N2 books“ conveys?
Is it, though?
I totally believe that newer users are happy with the labels „N5“ and „N4“ as these probably indicate „you can read this“ and also the others as they indicate „too hard, stay away“ - and that’s not a lie. I‘m just wondering how much more information this gives compared to the „Beginner“ etc. labels.
I mean, what is the difficulty scale, and how important is it? Larger number means harder, that’s how I see it. Everything else is just conveying a false sense of precision and accuracy in my opinion