I've made a draft of a howto:romanize wiki page, since we've been kind of needing it for a while, and since haづki doesn't seem to have gotten around to it, despite saying twice he'd do it.
It's based on my understanding of the current romanisation rules, referencing mainly forum #12961 and forum #31038.
The readability isn't all that, so if you can rewrite it to be clearer, please do. And of course, correct any mistakes, fill in any omissions, and add more illustrative examples.
Though I refer to revised Hepburn for anything not covered, I tried to make the rules complete enough to stand on their own.
Looks good. The only thing I've thought about that's not there is the particle へ, which is pronounced and should be written "e" and not "he" (as with particle は being "wa" and not "ha"). Don't think we have many examples of it anyway, though.
For Japanese names, these rules take precedence over any officially provided spelling.
Nrvnqsr_Chaos, anyone? I don't see why our guidelines should trump official spelling, and I can't think of any examples of this rule in action.
For foreign words like "lyrical", follow the spelling of their source language.
Including non-English loan words? Romanizing ランドセル as 'Raenzel' and テーゼ as 'These' strikes me as unnatural.
おお→oo
This is strictly personal preference, but I actually prefer 'oh' here, as well as 'Tohno' for 遠野 and 'Senoh' for 妹尾. It still unambiguously represents the underlying spelling, encourages proper pronunciation even for those not literate in Japanese, and just plain looks better.
Also, a semi-related topic that could fit into this article is how to romanize kanji-only artist names, which often have multiple plausible readings and no indication of which one is correct. Given, my technique for this is 'slap it in ENAMDICT and pick a common-sounding one', but even that might help any enterprising newcomers who want to upload stuff from a new Pixiv artist, but can't read Japanese.
Nrvnqsr_Chaos, anyone? I don't see why our guidelines should trump official spelling, and I can't think of any examples of this rule in action.
Doesn't this fall under the "non-japanese and clearly made up" exception?
Also, a semi-related topic that could fit into this article is how to romanize kanji-only artist names, which often have multiple plausible readings and no indication of which one is correct. Given, my technique for this is 'slap it in ENAMDICT and pick a common-sounding one', but even that might help any enterprising newcomers who want to upload stuff from a new Pixiv artist, but can't read Japanese.
I still say that we should explicitly make a "if you aren't 100% sure on an artist name, MAKE A FORUM THREAD" rule.
glasnost said: Nrvnqsr_Chaos, anyone? I don't see why our guidelines should trump official spelling, and I can't think of any examples of this rule in action.
That's not a Japanese name, now is it? This rule is put into action all the time, to varying levels of gnashing of teeth, but for artists and obscure characters it's fairly routine.
Including non-English loan words? Romanizing ランドセル as 'Raenzel' and テーゼ as 'These' strikes me as unnatural.
Yes, certainly. E.g. forum #23772. I thought there was a thread arguing using Japanese spelling for things like ランドセル, where the word has been adopted by the Japanese to the degree that it has it's own meaning not accurately expressed by the original word, but I can't seem to find it.
Fencedude said: Doesn't this fall under the "non-japanese and clearly made up" exception?
zatchii said: That's not a Japanese name, now is it?
Derp. It seems I somehow managed to overlook that exception while trying to think of an example. That said, I do still have a bit of a problem with this policy. I guess I can understand standardizing romanization of Japanese character names on our end, since the Japanese themselves don't seem to put much thought into 'official' romanizations, and often contradict themselves. An artist name, on the other hand, would be the last thing I'd want to change; if the artist has gone to the trouble of romanizing their own name in a consistent way, that's pretty clearly how they want to be identified, regardless of how 'creative' their romanization is.
Huh. I guess if the topic's already been beaten to death, I'll go along with that; I don't really have any substantiative objection, other than "it feels weird". Words like ランドセル should definitely use the Japanese spelling, though; Japanese has lots of false friend loan words, and romanizing アイス as 'ice' or カンニング as 'cunning' is just plain misleading.
I just read over the current version (as of this posting) and I think it looks good.
As for loan words, they can be taken on a case by case basis but it is pretty common for us to use the source language spelling. I still believe this exchange is a reasonable guideline:
jxh2154 said:
0xCCBA696 said: I think this should be decided on the basis of semantic shift from the original point of loan. ... In Japanese, アルバイト refers to a specifically part-time job, whereas in German, "arbeit" simply means "work". A significant semantic shift has occurred. So I think I'd go with "arubaito".
That sounds pretty reasonable to me, thanks. I'll try to keep that in mind if this comes up again.
And we do not accept artists' own romanizations of Japanese names. That won't change, and we've already done literally hundreds of aliases in line with our policy. What the artist is thought to want just isn't taken into consideration, to put it bluntly.
The collation thing may have been a foolish move from me, as I have been on the beginner level with hiragana and katakana for some years now. It was a by-product of my attempt to check whether any important basic guidelines were missing. (In my opinion, at least adding "ち→chi" was a good idea.)
I sincerely hope I fixed more than ruined, as I tried really hard. I humbly ask that somebody who knows these things check and correct the current version.
Thanks for the hard work, Katajanmarja. I read it over and I think it looks pretty good. I don't think it's necessary to write "(han)dakuten" every time, though, when you're only providing dakuten examples.
*laughter* I am so happy if that is the biggest issue there. Frankly speaking, I did not even check the difference between dakuten and handakuten, just tried to give the groups correct names.
Oh, and one further idea: something very basic, like akka, could be good for the first example in the gemination (double consonants) bit. After all, ch and sh are digraph spellings in English, while n is a rather special consonant in Japanese.
