Do you trust Natively levels?

I’m curious whether people in general trust Natively levels or are more skeptical. I’m normally skeptical of the levels for books with only gradings from a few users, but otherwise I think they tend to be alright. But then I run into stuff like this that shakes my confidence… 姫乃ちゃんに恋はまだ早い1 | L16 is level 16 and has gradings from 8 users, which is a decent sample size. Yet this manga does not provide furigana for all text. I haven’t read it (and since I don’t rely on furigana I wouldn’t trust my judgement anyway), but I find it hard to believe that really any manga that doesn’t provide full furigana could be level 16.

Anyway, what do you think in general and/or with this specific situation?


Furigana does not really matter with how good ocr has become.

I do trust the natively ratings (if there are more than a couple users rating),
but more in broad terms.
A 30 is gonna be more difficult than a 20.
But is a 28 actually easier than a 29 or 30? Depends on the person and what they read before :man_shrugging:


When I read a bunch of novels by Keigo Higashino, a mystery author for adults, one of the easier ones came out as Level 25, which is grossly underrated relative to other novels on this site. It’s probably more of a Level 30? (I don’t really judge by the JLPT, which the levels are pegged at, but by relative to other novels - I haven’t sat the JLPT lol.) Either way, it did not belong next to 魔女の宅急便 . If there are few ratings, they might be skewed by what the submitter initially put the level as + what the rater already read. I do trust the levels if there are more ratings, but I judge it relative to what I’ve read on the site and what I can infer is a Level 30 etc.


It’s a bit of a roller coaster to me.

There’s a bit of time where I think it’s fine, then I get to a streak where I go “There’s no way this is LXX”.

In any case, I don’t guide myself solely by the level to judge a difficulty, I also research the way it’s written/the setting and read a sample if it’s available if I’m going to go into one.

Unless it’s something I know I’m going to be able to handle before hand, like a Tsubasa Bunko or a Manga of something I already read the LN.


I never payed any attention to the natively level. In my case I read what I want no matter how hard it is.


Do most people really use OCR though? I would think not given you need some technical knowledge, but it also only became popular/good after I no longer needed furigana, so I could be wrong.

Yeah… I’m thinking about this now because WaniKani users just started a new intermediate manga book club, with the main goal being L27+ manga. But I don’t know how strict the club should be since there’s always variance based on who read/graded the manga.


At least in the spheres I frequent mokuro is the goto.
I have pushed it onto multiple beginners myself :sweat_smile:

1 Like

I couldn’t imagine doing all my reading on a computer. :joy: I’ve heard there are ways to get this kind of thing on Android-based e-readers, but then that’s getting even more complicated. Glad I don’t need that at this point. :sweat_smile:


As an outsider just giving the opinion, I’d say it shouldn’t be too strict and just be used as a reference.

The thing is always this rating bias for example:

  • How long has been it since you have read what you get prompted to grade, the system afaik doesn’t take this into account.

  • People have different levels of “Understanding”. If you go with something that it’s for example related to Shogi or Go, do you include into your ratings understanding the move names on it or not? Because if you do, I’d argue some of those series shouldn’t be as low as they are.

And some other things I might be missing too


I generally don’t trust the difficultly level if it has fewer than 3-4 people grading it. Some people rate things super easy or super hard because it is so far below/above their level that they have a hard time identifying the actual difficulty. If more people grade a book/video then it sort of evens out and is much more reliable.

As someone currently reading through 姫乃ちゃんに恋はまだ早い | L16 I can say that the rating is actually pretty accurate to the language used in terms of grammar and vocab.

As for the lack of furigana, that doesn’t seem to affect the rating. I (and I think most people on this site) grade without taking furigana into effect. And many N4 (L13-L19) manga series do not have furigana. I checked through them all and found that about 3/7 do not have furigana.

Honestly, I am glad that people seem to grade without taking furigana into effect. As someone who it pretty bad with kanji at the moment, I can slowly work my way up the difficulty levels using content without furigana. This lets me find content that should be just above my current kanji level which helps me to improve without getting frustrated. If the presence of furigana affected the rating, then that would muddy the waters and make it more difficult to find material that is appropriate.


It’s always been giving me a headache reading how to get it to run and prepare :sweat_smile:.

I use the vastly inferior but more lazy Capture2Text for when I want to look up a word and I don’t feel like playing with the kanji recognition or stroke orders on my electronic dictionary.

I have a bit of a different perspective since I’m only using beta languages (Korean and Spanish). For Korean, I look at how many people have graded it and especially what the individual gradings actually are. Consistency is key, and if the gradings are kinda wacky (similar to L21 and similar to L28, with a small number of gradings) then that gives me an indication that maybe it’s not the most accurate level.

I also look at who graded it, and someone that’s read a lot of books is probably going to be a good judge of grading, and thus the level would most likely make more sense, even with only 1-2 people grading. Sometimes it’s just one person that grades, but they only read ~5 books and have only graded these in relation to one another, and nobody else has read them. That’s the gradings I trust the least.

For Spanish, I’m considering broader differences in levels. L30, L35, L40, L45, that I follow, but anything more precise than that I don’t really consider. The number of gradings is still too small imo for it to have small level differences that make sense for most books right now.

So I’m always taking the gradings with a grain of salt, even moreso for Spanish. I feel like the recent benchmark creation for Korean has rendered Korean levels a lot more accurate though. I still tend to try and read samples before buying a book, because you never know when you’ll stumble into a book with a completely nonsensical level. It’s happening less and less, but I’m still on the lookout for those nonetheless.


It’s great to hear your perspective on this! I went through WaniKani to learn a ton of kanji long before Natively became a thing, so it’s really hard for me to judge. I always assumed people ignoring furigana in gradings largely came from a WaniKani bias (since Natively got a lot of early users from WaniKani), so it’s good to hear that people ignoring furigana can have a positive effect too!


I’ve found that very low ratings tend to be suspect because there’s nothing lower than them to push back up. I also have a feeling that people who are reading below their levels can muck up the low scores a bit (I feel like I did this with low level videos a while back as I’m trying to increase my ability to watch anime raw).

I also think that except for the most reviewed items that most levels are something more like +/- 2. I basically treat the bands as mostly the same and will probably be comfortable in purple, orange is a stretch, and we don’t mess around with green/teal.

I DON’T think that a level 25 novel is the same difficulty as a level 25 manga, though. I treat them as independent scales (even though they can technically be graded against each other and this shouldn’t be true).

This. I don’t read physical books anymore. Manga below my level I’ll read and just Google lens when I need a lookup. I’ll Mokuro anything that’s slightly below my level or higher.


This is a very good point. I bundle levels in a similar fashion even in the non beta JP language. Like for example in my mind a L22 and a L25 might feel almost similar in difficulty.

Probably if the level colors weren’t in 10 level intervals I’d be judging by colors rather than levels.

Your mileage might vary, but I haven’t noticed big differences in jumping within 3-4 levels, at least until the levels I have tackled, I’m not sure if this still applies on the harder stuff that I haven’t dared yet to venture into.

I try not to get the furigana come into an effect, but it probably still does, even if a bit.
Having to look up a word reading is not the same as having it right there, and being able to input it fast into a dicitonary.

It also depends on where I’m reading it, despite that it shouldn’t, and I try to compensate this when I’m grading, for example, if I’m reading a book without furigana and complicated words in an electronic device, I can just highlight, making it easier to look-up words.

If I’m reading this same book physically, then the lookup process becomes more difficult/tedious. Despite being the same difficulty, my first impression that I’m trying always to fight is that the difficulty should be the same.

While I do compensate when I’m grading, if I read a book physically X time ago, there’s a big chance, when I get prompted to compare in a grading I might have forgotten I read that in a physical medium, and only remain in my mind the difficulty I had with it.

/I’ll end the furigana tangent I have gotten into :sweat_smile:


As others mention above I don’t think furigana is that important to how difficult a text is (though I suppose I’m a little biased since I’ve been just reading whatever with or without furigana for this whole time now basically). It makes it less convenient to look up words, certainly, but I don’t personally count it as affecting the difficulty on the text itself as such. I don’t even use Mokuro or even OCR either for the most part, just a handwriting recogniser.

For me difficulty is based mostly in the amount of unknown vocabulary and complexity of sentences (and for manga, to a certain degree the difficulty of comprehension based on the strength of the comic storytelling etc).

That said, I do think the gradings usually sit within about a 3-4 level range, so something at 16 I would think could be anywhere for (for me) 14-18. Basically I only really think about them in terms of blocks of 3 or so (especially now at the higher levels where difficulty becomes a bit nebulous since my lack of vocabulary is more field-specific - there’s not functionally much difference between a difficult 26 and an easy 29, so to speak)

/edit also for my purposes novels are on a completely different track. I can motivate myself to fight a lvl 30 manga easy, but I still need to build stamina for novels, so even a level 25 right now is kicking my ass… though that could just be comparing 同じ夢 to 駄菓子屋 lol


Reading digitally, I’d be inclined to say furigana matters less because so much of it is unreadable anyway :stuck_out_tongue:


With Kaku OCR at least I reliably find myself fighting the OCR at least as much as I would just using a handwriting recogniser lol

Depends. If it’s manga, the quality is hit or miss and will depend on the device you are reading.

If it’s a novel, then you can make the text bigger and read it easily unless it’s one of those books that are basically images, which I despise with all my might.


If you wanna cry a bit, check out the preview for Probably the worst manga scans I’ve ever seen. (It makes me particularly sad because I probably would actually read this series if it had good digital quality.)

Here’s one screenshot as an example.