Status of Japanese requests

Mark Crispin mrc+ietf at Panda.COM
Sat Sep 26 07:18:00 CEST 2009

On Fri, 25 Sep 2009, John Cowan wrote:
> Well, straightening up the di/du is the main reason to distinguish
> between Nippon on the one hand and Hepburn and kunrei on the other.

But my point is that there are many variants or combinations that don't 
easily fit.

> Anyhow, even if conversion software doesn't map ee to ei, it *could*
> do so without knowing the exact romanization style.

I don't see how.  The difference between ee and ei result in different 
words in the written language even if they are homonyms in the spoken 
language.  Conversion software would have to have some AI-style context 

きれい (kirei) is "pretty".  切れ絵 (kiree) is a scrap of a picture.  OK, 
this is a rather contrived example but it was what came to mind.  I'm sure 
that one of the linguists could come up with something better.

> Just so.  Nippon's the only one that's mindlessly reversible.

Of the formally defined romanizations, yes.

Word processor romanization is mindlessly reversible (duh!), and doesn't 
completely fit into any of the three categories.  It's sort of a fusion 
albeit one that abolishes ambiguities.

IMHO, most people who write in romanization (whether by pen or keyboard) 
use a flavor of word processor romanization.  Long vowels are doubled, but 
use ei and ou following kana.  "shi" is preferred to "si", "ja" to "zya", 
etc.  First first processors in the 1980s, then personal computer input 
methods in more recent times, have shifted word processor romanization to 
the forefront.

That's what I do.  For the d-line, I personally prefer
 			 だ   ぢ   づ   で  ど ぢゃ ぢゅ ぢょ
 			---  ---  ---  --- --- ---- ---- ----
 	Mark		 da  jhi  tzu   de  do  jha  jhu  jho
 	Hepburn		 da   ji   zu   de  do   ja   ju   jo
 	Nihon		 da   di   du   de  do  dya  dyu  dyo
 	kunrei		 da   zi   zu   de  do  zya  zyu  zyo
since it corresponds to pronunciation (like Hepburn) but is mindlessly 
reversible (like Nihon).  Kunrei is a worst-case in both respects.  My 
forms also corresponds nicely to Hepburn's t-line:
 			 た   ち   つ   て   と ちゃ ちゅ ちょ
 			---  ---  ---  ---  --- ---- ---- ----
 	Mark/Hepburn	 ta  chi  tsu   te   to  cha  chu  cho
 	Nihon/kunrei	 ta   ti   tu   te   to  tya  tyu  tyo

Fortunately, the oddball d-line mora are all quite rare, although they do 
trip up people who spent too much time with romanization (such as Jorden's 
victims) without learning kana.

Romanization is a mess in Japanese.  I heard that it's far worse in 

Anyway, I'm not advocating any specific course of action other than to 
leave some expansion and not assume that a three-way Hepburn/Nihon/kunrei 
addresses Japanese romanization as it is actually used.

-- Mark --
Democracy is two wolves and a sheep deciding what to eat for lunch.
Liberty is a well-armed sheep contesting the vote.

More information about the Ietf-languages mailing list