I’ll be reading to the end as well. Might struggle or give up on some (more) comprehension, but I’ll stick with it. Goal is actually to try for more comprehension than I have lately.
2 Likes
True. I’m thinking a better ranking system would be to gather up all the books on Floflo and make three categories based on unique words, unique kanji, and avg characters per sentence. Average its ranking positions in all three categories to create a final difficulty value.
I’m not sure I wanna program that lol
4 Likes
Yeah, I saw them mention a weakness with unknown vocab, but I would’ve figured it should handle wikipedia well. Not so sure it is even very applicable to novels, in that case. Unfortunately I don’t have any e-books, or I would try it myself.
It does also just straight-up give the words per sentence, though, which I think is a bit more relevant (but harder to compute) than characters per sentence. I tried it out with a few wikipedia articles, and I can’t speak to the accuracy of the readability score, but it does seem to at least parse the number of words correctly. It actually counts ‘short-unit words’, so for example it splits:
格子系において
as
格子-系-に-おい-て
So it’s perfect for evaluating complexity.
Though flo-flo also has word parsing, right? @Raionus, no idea what would be involved to do that from your side, but maybe we can use that as a basis for a score?
—
Edit: Just saw Jreadability gave the first sentence of ハルヒ a readability score below 0. That puts it at significantly more difficult than transcriptions of the National Diet. This has proven to me the score’s validity, and no-one will convince me otherwise.
6 Likes
Aye, I got a score of -0.33 when I put in one sentence. I thought maybe it just wasn’t getting enough context to actually generate a score (the “Text Readability Level” is coming out as 測定不可 = uncalculable), but when I put three sentences in, it dropped to -0.39.
I believe this is simply because the score is outside expected ranges for the preset Readability Levels. It’s basically a level beyond Upper Advanced. Which I don’t think you could lucidly argue ハルヒ is.
For reference, the levels by score:
—
Edit: From the paper: “For example, an extremely short text that includes many kango in long sentences could produce a score less than 0.5.”
Which is pretty exactly what we have in the prologue. I suppose a fairer evaluation would need to include a larger sample from the book, but at this point, I feel like a more straightforward metric like avg words per sentence or unique word density (a la flo-flo) would make more sense to use.
5 Likes
Aw man and I wanted to read Shin Sekai Yori one day. Yikes. Though Hyouka sounds interesting as does Katakoi, but I should probably read things I’m already reading.
1 Like
Where does Konbini fall on that list, if you don’t mind?
2 Likes
Played around with the website a bit. Here’s what I get for the whole prologue of Haruhi vs the first 800 characters or so of Mishima’s Patriotism—might contain errors since I used Google Translate’s OCR to scan them, but seemed OK at a glance and should be enough for fun.
| Haruhi | Patriotism | |
| Text Readability Level | Lower Advanced | Lower Advanced |
| Difficult | Difficult | |
| Readability Score | 2.07 | 1.94 |
| Total Num of Sentences | 27 | 13 |
| Total Num of Word Tokens | 983 | 526 |
| Total Num of Lexical Types | 347 | 288 |
| Total Num of Characters | 1582 | 814 |
| Num of Words per Sentence | 36.41 | 40.46 |
This is taking into account the first three sentences of Haruhi, which are extra long, but also the first two paragraphs of Patriotism, which are written in literary Japanese.
4 Likes
35.6
That sounds intimidating, but I think that’s just a combination of long sentences + lots of kanji, which the test is biased for. The actual book has a unique vocabulary account like 10% higher than Kino so I don’t think it’s going to be that scary.
6 Likes
Huh, that’s really odd. I was expecting it to be around Kino’s numbers since it’s much closer in difficulty to that than to Haruhi.
2 Likes
Rejoice with trembling!
7 Likes
NOOoooo!!! *with trembling*
3 Likes
D: trembles
2 Likes
Y’all are missing the rejoicing bit!
6 Likes
I guess I’ll try to do that part(with emotes)
2 Likes
Messed around with different parts of Haruhi Vol1 and I can’t find any part that’s anywhere nearly as hard as the prologue. The first ~1700 chars of the first chapter is a 2.63, another 1900 char selection I made from the first chapter (avoiding quotes since they’re often easier) was 3.11, and a section that was mostly quotes was 3.72. The end of the epilogue is 2.29 (but the mostly-quotes first section of it is 4.5)
4 Likes
Yaaaaaaayyyyyyyyy (even though I haven’t even started Chapter 1 yet hahahhaha)
Edit: trembles 
3 Likes
Gosh and here I didn’t think Haruhi was that bad. Once you get through all the fantasy and SF vocab that is.
Well… of the people who filled out the Google Doc for the reading aloud on discord… it looks like 3 of you are all in the same time zone (Japan) + 1 about 8 hours off…
And the rest of us are -3 to -6 GMT. So I propose two small groups… and @Kyayna picks the one that fits the schedule better (for the offset timezone).
@StarLi @LucasDesu @Belerith = Japan Group (You three can decide your own time)
@Kyayna — choose up or down ??? Or wait and see what times we pick, then choose?
Myself, @Time, and @sigolino = Western Group (We can decide another time)… (I prefer Saturday some time…)
Does that work for everyone who was interested? I only tagged those of you who put a time down.
2 Likes
I made a voice chat channel on the discord server for you all to use.
6 Likes