Donmai

Tag Alias: houdzuki_akane

Posted under General

葉月 said:
What we don't have:

  • "oh" for おう. It's a totally retarded, self-inconsistent and completely unjustified aberration introduced by God-knows-whom (I suspect clueless US fansub typesetters[1] played a big role here) and inexplicably picked up.

It's still used in a lot of tags, maybe the most popular ones being tohno_akiha and tohno_shiki.
Edit: Never mind, I'm thinking of おお.

葉月 said:
What? Why? おう is by far the more common one, おお occurs in a couple kanji, most of which only have it in name readings. For practical purposes, it's limited to 大, 多 and not much else.

Consistency would dictate that one sticks to either arguing for a scheme that preserves the original spelling or one that reflects pronunciation. Meaning not proposing some weird hybrid scheme that consolidates homophones and then spells them out to reflect the original characters.

Updated

LaC said:
Because it's a long "o", not an "ou". "ou" is like "si" or "zi": for example, しよう is "siyou" if you follow kana, "shiyoo" if you follow pronunciation.

zatchii said:
Consistency would dictate that one sticks to either arguing for a scheme that preserves the original spelling or one that reflects pronunciation.

Please define "reflect pronunciation". If you go by English rules, "oo" would be phonetic spelling for う. If you go by, say, Polish, it'd be spelling for お・お, because we don't have long vowels, but do have glottal stops. Spelling is *always* arbitrary at some level, there's no law given by $deity to say that glyph X corresponds to sound Y. Nothing intrinsic says that "shi" should correspond to し, especially given that in English it'd mark the sound Polish spells "sz" (and amusingly enough, Hungarian spells "s", reserving "sz" instead for the sound we mark with "s"); it marks a sound that doesn't even exist in English to begin with.

What I'm saying is that you're looking at the wrong problem. It's not "spell kana correctly", because the only way to do it is to use kana. Instead, what you really want is to transcribe kana in a way that's aesthetically most pleasing[1] and least confusing. In the case of "ou" vs. "oo", you're looking at a subproblem of the above, namely "how to mark long vowels". We ditched macrons for a number of reasons, and instead opt for using "u" as a length marker for "o" and "u" (this is how you should look at "ou"[2]). This reflects the Japanese system and matches wapuro romaji, which, in absence of other, important reasons to do otherwise, is a desirable property. Additionally, it does away with the complaint commonly voiced by English speakers about "oo", namely "it makes me think it should be pronounced う".

[1] Which, obviously, is an arbitrary metric again. But modified hepburn generally closely matches the aesthetics of English-speaking audience, which danbooru is.
[2] Ie. おう == お + long marker, うう == う + long marker. This is logically matches long え, for which the marker is い, giving えい.

In pronunciation, we have vowel quality and quantity (length) as independent axes: ā is the same sound as ă, only longer; so is ō to ŏ, etc.. Following pronunciation means reflecting this structural property in the way you represent the sounds.

Examples of consistent systems:
- a,e,i,o,u indicate quality; long vowels are marked with a macron: a,ā, e,ē, i,ī, o,ō, u,ū.
- a,e,i,o,u indicate quality; reduplication indicates long sound: a,aa, e,ee, i,ii, o,oo, u,uu (bonus points: it's also consistent with the method used to mark long consonants, eg "ss").

The system you propose is a,aa, e,ei, i,ii, o,ou, u,uu. This sucks because a,e,i,o,u no longer match vowel quality accurately (eg, sometimes "u" is actually the same sound as "o"), and quantity is no longer represented consistently. Also, you can no longer represent the difference between ō (as in 勉強) and ou (as in 思う).

Of course, if you make the graphical unit larger and larger, you can say anything is pronunciation-based. Heck, English spelling can be pronunciation-based too if you consider entire words (better yet, sentences) as the base unit: after all, we all know how to match the word "women" with its sound. Still, you should be able to appreciate the difference in how finely the written form matches the sound.

We have and should keep a system that is a mixture of "common use" and consistency.

Common use on 'ou' is more than enough to rule out converting it all to 'oo', which would marginalize Danbooru more than any romanization choice we've ever made or could ever make. I care much more about keeping 'ou' than what we do with the less important (because far less frequent) oo/oh question, though I prefer 'oo' if I had to choose.

People are getting really hung up on the wrong type of "consistency". There is consistency across the entire romanization scheme. This sounds great at first but is not as admirable a goal as it sounds. For every good decision that comes of it, there will be some awkward decision that comes along for the ride (like 'oo' for おう ...Toohoo?).

The more practical type of consistency is to be consistent with each individual choice. Is おう going to be 'ou'? Fine, then always do that. It doesn't have to be a perfect conceptual fit with what is done in every other case (it does not require づ to be 'du' or 'dzu'), if embracing one thing requires you to embrace a problem elsewhere.

So you're all free to continue debating this but I, at the least, will be opposing any radical romanization changes. This is still a website that has to be used by a large number of very different people as intuitively as is reasonable, and *absolute consistency is not always intuitive*. Please keep that in mind.

葉月 said:
...

Right, but if you're not trying to get something that indicates the correct pronunciation, there's no reason to merge おう and おお, which after all are distinct in the original written Japanese.

I'm not trying to "spell kana correctly", the goal is to have an useful representation of the name in ASCII, that is readable for people who can't read squiggles. Somewhat mirroring the pronunciation is nice and makes the scheme easier to remember. Being one-to-one makes it just as useful for people who know Japanese as the original kana would be (e.g., for googling). If we don't care about people who can't read kana or don't have a Japanese IME, we can just use kana, of course.

While 'oo' can represent a lot of different sounds in English, several of them う-ish, 'ou' just hints at other wrong pronounciations, so it doesn't do away with the complaint at all. It seems to me that they're both equally rarely used to represent おお-like sounds. 'Oh' seems better in that regard, but you don't sound like a huge fan of it.

About defining "reflect pronunciation"; Ideally, it'd mean "being such that an average reader, employing the language's standard pronunciation rules, would read it as intended."
Of course, English being what it is, I'm not sure if there actually exists any string of letters with an unambiguous reading. I guess the best you can hope for is "an average reader, upon being told how the word is pronounced, would say “Yeah, that sounds plausible.”"

zatchii said:
While 'oo' can represent a lot of different sounds in English, several of them う-ish, 'ou' just hints at other wrong pronounciations, so it doesn't do away with the complaint at all. It seems to me that they're both equally rarely used to represent おお-like sounds. 'Oh' seems better in that regard, but you don't sound like a huge fan of it.

If you don't know how to pronounce Japanese vowels in the first place you are likely to botch things up. For example, someone who's not paying attention will easily read "gyaggu manga" as "gyaggu meinga" or (horror of horrors) "loli" as "lalli".

I redact my particular dzu/zu complaint here, which I really just posted to bring up undiscussed Romanization issues. The existing system is to me quite sensible and there is no need for any major revisions. I'd like to suggest to rq that haduki's rules be written up in the wiki and linked to somewhere.

jxh2154 said:
Let's keep what we're used to.

Do you think that's a fair representation of your argument? As far as I can tell, that's all it boils down to. I don't see any other reason for preferring "Touhou" over "Toohoo".

zatchii said:
Being one-to-one makes it just as useful for people who know Japanese as the original kana would be (e.g., for googling).

This is indeed a big part of why we want to represent long vowels in the first place.

Of course, English being what it is

Wait, why are you bringing English into this? Obviously English spelling is too broken to take it into consideration (although I'll concede that spelling 東方 "Toe-hoe" would be amusing).

LaC said: Do you think that's a fair representation of your argument?

Only if you didn't read my post. Please stop vastly oversimplifying arguments you don't agree with, because it really makes it hard to discuss anything if it basically sounds like you're not interested in listening. A "fair representation" of my argument would be... well, exactly what I had posted before.

And what would be so bad about "sticking to what we know" anyway, if it works, and the alternatives would cause problems? Doing something like "Toohoo", "Suigintoo", "Mooryoo no Hako", "Mahoo Shoojo Lyrical Nanoha" etc would just be a big "fuck you" to the userbase of Danbooru, serving no end but intellectual masturbation for the heck of it.

Updated

jxh2154 said:
serving no end but intellectual masturbation for the heck of it.

This is the point where I make a joke about not seeing what the big deal about having more masturbation on Danbooru is.

But I think I shall refrain.

jxh2154 said:
Only if you didn't read my post. Please stop vastly oversimplifying arguments you don't agree with, because it really makes it hard to discuss anything if it basically sounds like you're not interested in listening.

Ok, but I don't see what that has to do with the situation at hand. I never said that "let's keep what we're used to" was a bad idea.

But fine, fine. Would "Let's keep what [we think] the internet anime community is used to" be a fair representation of your position, then? If not, please point out what I'm missing.

And what would be so bad about "sticking to what we know" anyway, if it works, and the alternatives would cause problems? Doing something like "Toohoo", "Suigintoo", "Mooryoo no Hako", "Mahoo Shoojo Lyrical Nanoha" etc would just be a big "fuck you" to the userbase of Danbooru, serving no end but intellectual masturbation for the heck of it.

That sounds so thoroughly appropriate and fitting that it's hard to believe you're not arguing in favor of the switch! But personally, I'm still discussing the theoretical merits of various approaches. I'm not arguing that we should change all those tags yet.

1 2