Monday 10 February 2003

Just make sure you spell it incorrectly

I’d completely forgotten there were different romanization systems for Japanese until Bertil Wennergren mentioned it in a comment. As a foreigner, I’m familiar with the Hepburn/Hyōjun system though I’ve always tried as much as possible to avoid rōmaji, believing that relying on transliterated Japanese makes it more difficult to read the actual language. Even so, since all my kanji dictionaries contain rōmaji, it’s impossible to avoid.

The entry on rōmaji in my Kodansha Illustrated Encyclopedia of Japan notes that the first romanization system, based on Portuguese, was developed by Jesuit missionaries in the 16th century. The Hepburn system, named after a Philadelphia medical missionary who arrived in Japan in 1859 was later modified and expanded into the Hepburn/Hyōjun system while the main alternative still in current use, the Kunrei system, has its origins in the Nippon system developed by Tanakadate Aikitsu, a physicist and professor at Tokyo University. The encyclopedia entry explains:

The Kunrei system remains in use for grade school textbooks and the National Diet [parliament] Library, among others, while the Foreign Ministry and most Japanese-English dictionaries continue to use the Hyōjun system.

The differences between the systems are shown in the illustration below:

Hepburn/Hyojun, Kunrei, and Nippon Romanization Systems for Japanese

A Google search on “rōmaji” turned up a fascinating article by Andrew Horvat titled The Romaji Conundrum in which he points out that although until recently the differing romanization systems were only of interest to educators, “these days, however, the party most concerned with Japan’s romaji chaos is the International Standards Organization which has put Japan on notice to come up with a single, rational, unified system.”

Why? Email addresses.

Horvat uses the example of a Japanese man whose given name would be transliterated as “Jun’ichi” (using the Hepburn/Hyōjun system). The Kunrei system would render the same name as “Zyun’iti” but the computer engineers who assigned his email address mixed up the two systems and assigned him the name “Jyunichi”. As he told Horvat, “if you don’t spell my name incorrectly you won’t be able to reach me.”

If past precedent is any guide, neither the ISO nor anyone else should hold their breaths. Anarchy is likely to reign for a long time to come. Instead, it might be an idea for those with an abiding interest in Japan to learn that “rōmaji” (the use of Latin letters to write Japanese) can be spelled as “roumaji,” “roomazi,” “roomadi” and any number of other variations.

Needless to say, the Japanese bureaucrats charged with deciding on a uniform standard are said to favor the Kunrei system even though young Japanese—those most likely to own computers and have email addresses—feel more at home with the Hepburn system.

Permalink | Technorati


I hate that, in learning character-based languages in school, you're taught the romanized version first (this is the case for everyone I've talked to, at least) and then the characters. And then depending on the professor's preference, you might have to learn multiple different romanization system...

In learning Chinese, I learned Pinyin in China (thankfully the central government has stepped in and mandate a uniform romanization for the mainland, making things a bit simpler), then came back to the States and had to learn Wade-Giles for my first semester Chinese class. Then that prof retired, and I had to learn Yale for the new guy! I spent more time learning new romanizations for characters I already knew than I would like to think about.

Posted by John on 11 February 2003 (Comment Permalink)

I hated that too, John, though my experience obviously wasn't nearly as pointless and unpleasant as yours. Interestingly, one of the most popular introductory books -- Japanese for Busy People -- now comes in a kana version as well as a romaji version.

Posted by Jonathon on 11 February 2003 (Comment Permalink)

I started out with the Jordan book in college, which is apparently reproduced from a typewritten source. Obviously no kana or kanji. But it was in kunrei-shiki (huzi, titibu).

Posted by Adam Rice on 11 February 2003 (Comment Permalink)

Golly, Adam! What a trip down memory lane. I made a couple of failed attempts to learn Japanese before finally knuckling down to the task. I'd totally forgotten the Jordan book (which was the first I used) but I've just found Alfonso & Niimi's Japanese: A Basic Course on my bookshelf and it's also reproduced from a typewritten source but using the modified Hepburn system.

Posted by Jonathon on 11 February 2003 (Comment Permalink)

You're complainging about Japanese and Chinese? It's far worse in Korean: almost pure anarchy. A new official system much every year or so, and hardly any of the systems make any sense. Searcing for Korean names in Latin script on the web is a nightmare. The names can be spelt in more ways that one can think of.

Posted by Bertilo on 11 February 2003 (Comment Permalink)

Then there are the various wapuro based influences on romaji systems, and the kana influence. I think, if I remember my systems right, the "ou" for long o isn't in any of the official ones, though it follows the kana spelling. Yet I see it a lot, and I use it. (I'm woefully inconsistant with my romanization and must work on that. How do you get the macrons on your o's? I've had to use a special font for that.)

It was odd seeing the /kwa/ and /gwa/ syllables on the Nippon side of the chart: I thought those syllables died out long ago (being syllables adopted from Chinese anyway), certainly before Tanakadate Aikitsu. Either my understanding of that process is wrong, or perhaps its an archaism in the system: my money would definitely be on the former, however.

Posted by Kristina on 17 February 2003 (Comment Permalink)

You're right about the wapuro influence, Kristina. Horvat mentions it in his article. I'd never heard of the 'kwa' and 'gwa' syllables until I saw them in the Kodansha encyclopedia.

As for the macrons, I use the following HTML character entities:
ā ē ī ō ū

You can copy them and use them in your own blog entries.

Posted by Jonathon on 17 February 2003 (Comment Permalink)

I found the passage I was reading today, which I'll type out for you. The source is Masayoshi Shibatani, The Languages of Japan (Cambridge Language Surveys series), page 203. The preceding paragraph involves changes during the Muromachi period.

"The borrowing of a great numver of Chinese words introduced the new syllable types [kwa] and [gwa] to Japanese. They were incorporated into Japanese and enjoyed their stable status within the Sino-Japanese lexical stratum until the seventeeth century, when they merged with the plain [ka] and [ga] in the then standard language of Kyoto. However, the opposition between labialized velars and plain velars is preserved in some parts of the Kyoto area and in such peripheral areas as parts of Tohoku, the majority of the Kyushu region, and a few other spots in the rest of the mainland. In those areas, the standard forms /kazi/ 'fire' and /kazi/ 'house work' are distinguished as [kwadzhi] and [kadzhi]." (I'm afraid I had to use zh for an IPA character that's a voiced palatal fricative.)

The fact that the Nippon system, though, has room for such dialect forms in its writing.... That's very interesting. Particularly if it was designed that way at the time. I would have expected it to be very standard-Japanese specific, but instead it has this room for the periphery....

Posted by Kristina on 17 February 2003 (Comment Permalink)

I printed an article "Proposal of Expanded Hepburn System" in "NIHONGO Gaku", which you can read at

Will you please read it.

Posted by KAMINISHI Toshio on 30 June 2003 (Comment Permalink)

I printed an article "Proposal of Expanded Hepburn System" in "NIHONGO Gaku", which you can read at

Will you please read it.

Posted by KAMINISHI Toshio on 30 June 2003 (Comment Permalink)

The PDF file of my proposal of Expanded Hepburn system at does not have multibyte characters embedded.

The proposal of `kakuchou hebonshiki' is written to put an end to the rohmazhi (my transliteration) chaos. I don't say everything goes well, but believe it is worth your consideration.
In kakuchou hebonshiki, `ji' stands only for the voiced counterpart of `chi', and that of `shi' is spelled `zhi'. The glottal stop is defined as `t' to cover the word-final position. The assimilated form usually takes the same sound or letter with the following letter, but when followed by `ch' it is `t'; when followed by `j', it is `d', and when followed by `h', it is `k'. And it is of course transliteration system. So there is no need of duration mark. `Ombiki' in katakana is transcribed by `h'.

Posted by KAMINISHI Toshio on 19 July 2003 (Comment Permalink)

The pdf file at was replaced with the new one thrice in size, with embedded multibyte characters.
Much obliged to the administrator of the site.

Posted by KAMINISHI Toshio on 21 July 2003 (Comment Permalink)

can you please e.mail the proper way to write adam in chinese writing a.s.a.p

Posted by niall on 21 November 2003 (Comment Permalink)

This discussion is now closed. My thanks to everyone who contributed.

© Copyright 2007 Jonathon Delacour