Danbooru

[Wiki Rewrite] howto:romanize

Posted under General

howto:romanize

Given a recent discussion over the appropriateness of standardized tags (and the page's inherent conflict with artist/copyright/character names that are localized), it most likely deserves some well-needed revision.

The most recent change I made is tentative; that being, I wrote it conforming to what I believe is most representative of the status quo regarding romanized tags and their normalized use.

Since things are subject to change, @nonamethanks brought it up in a chat with me whether the following clause even applied anymore:

  • For Japanese names, the Danbooru standard takes precedence over any officially provided spelling

This, in particular, is seen a lot less nowadays, even among some of the copyrights and characters with the most volatile inconsistencies in romanization e.g. mash kyrielight and a number of characters in the granblue fantasy franchise.

If there are any suggestions or corrections to be made, please share them.

Updated

Mikaeri said:

This, in particular, is seen a lot less nowadays, even among some of the copyrights and characters with the most volatile inconsistencies in romanization e.g. mash kyrielight and a number of characters in the granblue fantasy franchise.

In the case of Granblue and F/GO, it's Japanese writers attempting to create Western-sounding names, so an official translation is better than nothing. At least the official translation team may have had some corroboration with the JP writers, as opposed to the stab-in-the-dark with fan translations. It's the difference between "George Smith" and "Jouji Sumisu."

I think in recent years the landscape has changed enough that publishers are usually fairly consistent with their official translations/localizations, which is why it didn't really surprise me at first that we finally adopted the usage of those names as tags.

The way I see it, while it is true that oftentimes it would be a good idea to use the usual romanization standards currently provided in the wikipage, if such official translations or officially provided readings (given a degree of correctness) are all of the three I've listed:

  • Consistent (thus subject to little internal debate)
  • Prevalent (widespread and common usage)
  • Significant

Then it would warrant usage of the official localized or translated name as a tag, even while under semantic duress -- にこ→nico, 絵里→eli (both written in kanji/kana that isn't usually read as such, but is valid enough).

Not to forget that at any point in time is an artist allowed to provide the official reading of their name or adjust an existing one, for whatever reason. They might see they have western fans, and/or they might want to make their work easier to find. That's further reason to believe that there should be much more flexibility in the guidelines currently provided.

I agree that it's preferable to use the official name over our own standard when it's prevalent enough, if only because the purpose of tagging is to aid search, so using something nobody would ever search for just because it's "technically correct" would be pretty pedantic.

I'm not talking about alternate names or anything, just romanizations where official ones are well known but using a nonstandard version (niconico is another example) or one that would be ridiculous to correct (tokyo -> toukyou).

...one that would be ridiculous to correct (tokyo -> toukyou).

God forbid we start applying the same standard to named places. I'm pretty sure people would make dirty faces reading and using things like Toukyou and Oosaka (even though they're "technically correct" 😒).

For Japanese names, the Danbooru standard takes precedence over any officially provided spelling

I have a feeling the quote is meant specifically to resolve the difference in romanizaning style that a native Japanese would learn compared to the Hepburn style that foreigners usually use (i.e. Inaduma vs Inazuma). That is, it is never meant to override uncommon spelling adopted by Japanese writers (i.e. nico, eli) which I doubt is derived from any established romanization style. Perhaps it should be rewritten to convey as such.

Regardless of intention though, I believe most people would favor the official name for convenience (unless it's clearly a romanization style problem i.e. Inaduma).

---

In practice, I doubt romanization in tagging would be much of an issue given that it mostly involves character names which tend to favor whatever is commonly being used by fans. Trying to standardize translations like senpai/sempai and Gensoukyou/Gensokyo though, is nigh impossible.

Vezral said:

I have a feeling the quote is meant specifically to resolve the difference in romanizaning style that a native Japanese would learn compared to the Hepburn style that foreigners usually use (i.e. Inaduma vs Inazuma). That is, it is never meant to override uncommon spelling adopted by Japanese writers (i.e. nico, eli) which I doubt is derived from any established romanization style. Perhaps it should be rewritten to convey as such.

Regardless of intention though, I believe most people would favor the official name for convenience (unless it's clearly a romanization style problem i.e. Inaduma).

---

In practice, I doubt romanization in tagging would be much of an issue given that it mostly involves character names which tend to favor whatever is commonly being used by fans. Trying to standardize translations like senpai/sempai and Gensoukyou/Gensokyo though, is nigh impossible.

That sentence primarly needs to be addressed/rewritten to avoid stuff like topic #15176 from repeating.

nonamethanks said:

That sentence primarly needs to be addressed/rewritten to avoid stuff like topic #15176 from repeating.

That is an interesting read; I feel danbooru romanization rules have gone too far when even Google is trying to correct your spelling of a character's name.

On that same note though, feline lump brought up a good example of Tohsaka/Toosaka. Would the changes under Mikaeri's suggestion meant Tohsaka would be favored since it's more prevalent and most likely the one commonly used by the writer?

Vezral said:

On that same note though, feline lump brought up a good example of Tohsaka/Toosaka. Would the changes under Mikaeri's suggestion meant Tohsaka would be favored since it's more prevalent and most likely the one commonly used by the writer?

Indeed.

The long o → oh romanization rule is part of what's known as 'passport romaji'. Assumably, it's an adoption meant to convenience the pronunciation of Japanese names by foreign users who don't speak their language. An unfamiliar user may accidentally pronounce "Toosaka" as something like "two-sah-ka" versus pronouncing it when it's written as "oh", for example. Hepburn already does this, but this is just a furthered rule change.

I don't disagree with most of the proposals in the wikipage, but it is a bit strict (which is what we're addressing). Waapuro has long since been suited for western anime fans because of its ease of use, and admittedly it is by far the most proper romanization system we can use for a site like this. But incidentally it can lead to problems, especially with tag conflicts, where two different artist names are read the same because of non-unique kana -> romaji mappings e.g. Chizuru (ちづる) == Chizuru (ちずる).

If one wants an extremely clear mapping of names though, and needs to pick a romanization system across the entirety of the tag space, in order from most "Japanese" to least "Japanese" would be:

  • Nihon-shiki -- practically all romanization rules are a 1-1 mapping, nothing is lost in a conversion, it's just somewhat annoying to read for most westerners which makes it unsuitable to use.
  • Kunrei-shiki -- close, but non-unique rules for 'zi' and 'zu'
  • Hepburn/Hepburn-style waapuro -- the one we're most acquainted with. There are numerous conflicts and variations, but they usually are accepted in the end goal of usability.

So, none of these are perfect, simply because of the complications of going from one alphabet to another.

Then, I suppose this is less a problem of what romanization system to use (waapuro is already the most convenient and most understandable), it's about what names we should go with and which aliasi should we propose in these unique cases. Since romanization (and translation/localization) can vary across copyrights, my usual rule is "whatever they (the copyright owners or publishers) use, we use." If one isn't given at the time of creation of a particular tag, we make a best attempt and then we can fix it afterwards if we need to.

EDIT: One thing I also forgot to mention, there are even unique cases where we ignore certain provisions altogether because some names are particularly difficult to provide romanizations for, e.g. ahhien (あっひぇん-> ahhi...e...n?) or makadamixa (まかだみぁ -> Macadami...a?).

Updated

My observation is that the current precedent boils down to two rules:

  • If the standard official romanization is different from Danbooru standards, and that difference is attributable to a difference in romanization standards, change it. For example, the Idolmaster Cinderella Girls series has consistent romanizations for each character, but reverses name order, uses Kunrei-shiki transcriptions for some syllables, and truncates long vowels (e.g. post #1940268), so a handful of characters have significantly different tags.
  • If the standard official romanization is different from Danbooru standards, and that difference exists for stylistic reasons, do not change it. Nico, Eli, Sonico, et al. These romanizations can be identified by their lack of correspondence to any major romanization system, excluding some little-known historical ones.

Changing either of these would necessitate a number of aliases on top of the new wiki edits. I'm not necessarily opposed to this (there have been complaints about ryuu (street fighter), for instance), but we should be fully aware of those consequences before diving into this.

feline_lump said:

My observation is that the current precedent boils down to two rules:

  • If the standard official romanization is different from Danbooru standards, and that difference is attributable to a difference in romanization standards, change it. For example, the Idolmaster Cinderella Girls series has consistent romanizations for each character, but reverses name order, uses Kunrei-shiki transcriptions for some syllables, and truncates long vowels (e.g. post #1940268), so a handful of characters have significantly different tags.
  • If the standard official romanization is different from Danbooru standards, and that difference exists for stylistic reasons, do not change it. Nico, Eli, Sonico, et al. These romanizations can be identified by their lack of correspondence to any major romanization system, excluding some little-known historical ones.

Changing either of these would necessitate a number of aliases on top of the new wiki edits. I'm not necessarily opposed to this (there have been complaints about ryuu (street fighter), for instance), but we should be fully aware of those consequences before diving into this.

I can generally agree with this. Not much else to say, but just wanted to address that if we do go with something like this, the first point has to apply consistently to all the characters within a given copyright or copyright group. It would suck if both our style and theirs was consistent, but we befuddle ourselves over what to use. Appropriate romanization usage guidelines would apply specifically within the domain of copyrights, and this could (and probably should) be indicated through their wikipage.

1