romanisation is driving me batty

[piranha]

there are several systems of romanisation for japanese; only 3 of which really matter these days. since i got my first "instruction" in it via manga and anime titles, my set is kinda fubared, because the people who romanise titles don't all use one system, and worse, many mix the systems.

i understand why they mix. the best-known romanisation for english speakers is hepburn (not after audrey or katherine; after a guy who published his dictionary in the late 19th century). at some point later that got revised, and the revised system is what pretty much all dictionaries and most modern instruction devised for english speakers is written in.

even revised hepburn unfortunately uses macrons to indicate syllabic n (n̄) and long vowels (ū, ō, ī). not a problem for printing books, but most people don't even know how to type these. and so hepburn gets modified -- but not just one way. some people write n', uu, oo, ii, some write n, uh, oh, ih, some write ou for ō, some i- for ī... it's a mess.

but it doesn't stop there, oh no! the japanese themselves weren't happy with hepburn; they wanted a system that more clearly indicates japanese sounds for japanese speakers. so they came up with their own: nihonshiki (日本式ローマ字), which has, of course, since also been revised as kunreishiki (訓令式ローマ字). it uses circumflexes for long vowels; a bit easier to type for europeans. but it also uses ti (chi in hepburn), syo (sho), tu (tsu), etc -- not exactly intuitive for english speakers, ergo it's not surprising that it hasn't really caught on over here. the japanese are sort of accepting that in international communications -- but not by adopting hepburn for themselves, no, by making certain explicitly allowed exceptions (such as "passport japanese").

for example, you can see 東京 (とうきょう) romanised like this:

Tokyo -- in most international references; this isn't any proper romanisation, and IMO sucks because it results in nobody pronouncing it properly (though i guess it beats Cologne for Köln.) US manga publishers also use this "method" for names. i can see why -- because english pronunciation is so fuxx0red, but i'm griping about it anyway.
Tōkyō -- original and revised hepburn, seen in dictionaries and english study materials
Tookyoo -- modified revised hepburn
Tôkyô -- nihonshiki and kunreishiki
Tohkyoh -- passport japanese, and germans use this (which is WHY it is passport japanese i bet)
Toukyou -- wa-puro, which corresponds well with kana spelling [1]

i've been considering adopting kunreishiki just to stick with the way the japanese do things (cause if i ever go over there, that's what i'll need to use). but for translating manga, that's not a good idea. at this point i am doing what everybody else seems to be doing -- rolling my own, which is an unholy mix of modified, revised hepburn, kunreishiki, and kana spelling, *sigh*.

don't even get me started on word separation or capitalisation in romanised japanese. or on reverse translation of katakana. i have opinions!

[1] since starting to type into apple's kotoeri (IME = input method editor), i can see that my unholy mix works well with wa-puro (ワープロローマ字). wa-puro stands for "word processor" and is the romanisation most people use to type japanese into computers these days (including the japanese, who are leaving their kana keyboards behind). it's quite forgiving in that it needs no macrons or circumflexes, and it accepts nihonshiki, kunreishiki, and most modified hepburn. adhering to kana spelling rules helps me with remembering vocab as well.

Comments

[holyschist]

This is really interesting! I wasn't clear on how many romanization systems there were for Japanese, and the differences between them.

Back when I studied Japanese, I didn't have a formal textbook or anything, and I ended up using something much like wa-puro. I vaguely recall there being some kind of transliteration/transcription distinction in romanization in general, where one represents the sounds/phonemes better, and the other replaces characters more or less 1:1. I've always had a preference for the latter, although it can definitely be a problem for pronunciation (see Russian" zdravstvuitye," in which half the consonants are not pronounced).

[hobbitbabe]

When I studied Japanese in classes, we used a textbook with all Romanisation and then had a separate book for learning kana and kanji. I had the impression that the method of Romanisation had been made up by those textbook writers (Jorden and Noda) for the ease of English-speaking language learners - which meant that it led us into knowing how to write things in kana and having good first guesses at pronunciation, but was not really accepted by any Japanese speakers or anyone who had learned differently.

It was like Tookyoo and sensee.

[piranha]

yeah, eleanor harz jorden invented her own system for her textbooks -- i have the first two. JSL, her system, has one big advantage over hepburn (which kunreishiki also has): it makes it easier to conjugate because you can always simply change the final vowel of the root mora, while in hepburn you have to sometimes change entire endings. that has definite merit. i am more dubious about her other contention: that hepburn is more likely to lead to sloppy, unjapanese pronunciation, because we're supposedly too familiar with the letters in english. maybe i am an exception because this isn't my first foreign language, but transliteration in a language with which i am already familiar has never kept me from learning even subtle nuances of the new language. besides, i don't see how that point even carries, because no romanisation can represent japanese properly; it's moot whether "hu" or "fu" is better, because the actual sound lies pretty much in between those, and the same goes for the other syllables in contention.

i tried to learn from her books, but her method and the way i learn don't mesh well. she believes that people should stick to listening/speaking for a long time. i see her point, but i learn much better if i start writing right away, especially in a language with a non-latin "alphabet"; i learn pronunciation tied to writing. ergo i am not really willing to stick with romanisation for as long as she does it; i'm mostly done with it now that i know the kana well, and it hinders/distracts me more than it helps me. she also dumps too much extra stuff on top: she use n̄ for the syllabic n (which feels WRONG to me because that n isn't a long sound like the vowels represented with macrons), she uses extra accents to indicate pitch (which i am dubious about), she adds an additional set of morae starting with ḡ for a nasal sound (which is totally non-japanese and throws the kana tables out of whack, and (like pitch) is not universal but a specialized dialect thing that's all too easy to get wrong). and -- a bit surprising for a series that prides itself on teaching the spoken language -- the practice dialogue is stiff and stilted, and the vocabulary is quite limited.

maybe worst of all, these are the most deadly boring textbooks i've seen since the 50's -- and i actually LIKE grammar! so yeah; too bad, because i can see she was a exceptional scholar. i'll keep the books around as a grammar reference, because they're outstanding in that regard.

[green_knight]

Oh, good, something to look forward to.

All of the iPhone sources I've used so far stick to Toukiyou. which I took me some time to connect to the city I know as Tokyo, but then I knew Peking before it became Beijing, and you adjust. (If you want awkward, try Aix-la-chapelle/Aachen. French and German names sometimes don't map very well.)

The problem I have with romanisations that use a long vocal (and I'm happy with those thanks to a passing acquaintance with Welsh) is that the the Hiragana does not *have* long vocals - so you're in effect writing things different in Hiragana than in Romanji and that completely confuses me because I then can't get one from the other and feel I've made a mistake (my grasp of Hiragana is still shaky around the edges. When I try to type something, I don't want to translate from 'word I know' into 'word computer recognises'. A one-to-one correspondence is much more useful.

[piranha]

"Toukiyou" -- seriously? that is just wrong. i can accept all the official romanisations because in their own way they are consistent and make sense. but there's no excuse for Toukiyou; there is no 'ki' in the word in japanese.

i know i have too much of a bug up my ass about keeping proper names intact as much as possible. :) in europe i accept it as perfectly reasonable to have different names for cities depending on language -- some of them adopted the name way back in roman times, and simply never changed it. that's at least partly true for aix-la-chapelle; aix comes from latin 'aquis', and the hot springs of what is aachen today used to be 'aquis granum'. somehow i get bent out of shape when north americans change original names, *snrk*. like, they have no reason to. but in fact they do have reason to, because speech and writing don't match, and usually the writing adjusts to the speech, and the speech tries to approach the native pronunciation. it's stupid of me to get hung up on that.

yeah, i prefer a 1-to-1 correspondence for typing as well, and i pretty much have it with the way i mix things now; except for ん, and there is really nothing one can do about that because it's an exception to the rest of the syllabary. oh and the small vowel kana, which one could type (there are special keywords, i think), but it's much faster to just type "kke" or "tta", etc. i can deal with those exception, but i rather keep them to a minimum. there are already 3 systems of foreign writing to handle.

btw, it's not "romanji". i see that error a lot -- the word is 'ローマ字' -- see? no 'n'.

[green_knight]

(Ok, I know romaji in principle, but my fingers want to stick an 'n' on the end of roman every time.)

You're right about Toukyou, I falsely quoted from memory *because it made perfect sense to me*.. (I also pronounce it /tokio/ rather than /tokyo/ which does not help.)

After a lot of frustration I have worked out where the problem lies, and I am not certain I will ever be able to get around this: to my brain, とうきよう (which has a 'ki' in it) *is exactly the same* as とうきょう, which does not.

My inability to correctly decipher Hiragana is starting to make a lot more sense to me.

[piranha]

AH! yes, it makes sense that you'd have trouble.

i have trouble with the small vowels too, because in some fonts (and especially in manga (vertical writing makes it even harder)), i can't always differentiate, and i just don't know enough vocabulary yet to get it from context. i sound words out now to help with that, because i have more listening vocab, and i'll recognize at least common words. but i don't understand why they don't always print the damn things much smaller.

[trixtah]

From the perspective of knowing something about Polynesian languages, I prefer a macron, although using one over a consonant just seems WRONG (although obviously appropriate for Japanese). If it is used consistently to indicate a long vowel instead of a short one, it's reliable, IMO.

Macrons rather than doubled-vowels clearly indicate when there is a single sound rather than two (important when there is the kind of glottal stop you get in Samoan, etc). Using a diphthong when it's a single vowel seems potentially inaccurate (since I believe the "o" in Tokyo is supposed to be an /ɔ/ rather than /oʊ/), as does the "oh" rendering.

The circumflexes seem slightly odd to me from having studied French, but if used consistently, they seem to be like macrons in their usefulness (slightly edged out IMO by macrons, since the latter are not used in a quite different way in another language). Difficult problem, though!