A Short History of Romaji

This post is part of a collection on Japanese Language Technology.

I recently created a tool called cutlet for converting Japanese to romaji. While working on cutlet I was surprised at the difficulty of finding a good guide to the difference between romaji systems, particularly their history and usage, so I wrote this up.

A Romaji Refresher

There are three relatively well-defined romaji systems. For a quick overview of some of their differences:

Japanese 新橋 お茶漬け 富士山
Hepburn Shinbashi (or Shimbashi) ochazuke Fujisan
Kunreisiki Sinbasi otyazuke Huzisan
Nipponsiki Sinbasi otyaduke Huzisan

Hepburn is the most widely used outside Japan, Kunreisiki is officially taught to children in Japan. Nipponsiki is similar to Kunreisiki but older. Technically Hepburn uses macrons for long vowels while Kunreisiki and Nipponsiki use circumflexes, but usage isn't consistent and there's no difference in meaning between them.

In recent years "wapuro" romaji, based on input when typing Japanese, are also relatively common, but it's not systematized (except in as far as programs either take or don't take certain sequences) and isn't used in any kind of formal writing.

Particle Comparison

As another refresher, here's how non-phonetic particles are represented in the various systems.

Hepburn wa wo e
Kunreisiki wa o e
Nipponsiki ha wo he

Hepburn is phonetic except for wo, Kunreisiki is strictly phonetic, and Nipponsiki is strictly orthographic.

Early Romaji

Japanese was first written in the Latin alphabet in the 1600s, mainly by Christian missionaries. It was used by the missionaries for recording and learning the language, and also used in texts they produced to try to spread Christianity. Since Christianity was later supressed in Japan, these forms of romaji never had wide adoption, though they influenced later efforts.

The Origins of the Hepburn System

In 1867, missionary James Curtis Hepburn published a Japanese-English dictionary in Yokohama. Naturally he needed some way to write Japanese words for people who couldn't read Japanese, so he devised a system of representing Japanese that reflected the pronunciation of English consonants. His system wasn't designed from the ground up, but developed based on his knowledge of existing systems by other missionaries as well as his own knowledge of Japanese.

Somewhat later, in 1885, an organization known as the 羅馬字会 (Romajikai) was established. Mainly a society of Japanese academics, it was interested in promoting the use of romaji as a replacement for kana and kanji. Some motivations for getting rid of native writing systems included easier international communication and improved domestic literacy. The principle founders were Ryoukichi Yatabe, a botanist and the first Japanese graduate of Cornell, and Masakazu Toyama, a graduate of the University of Michigan who later occupied a high post at what would eventually become Tokyo University.

The Romajikai published books to promote romaji, including this translation of Max and Moritz, titled "Wampaku Monogatari". This edition was reprinted in 1978. via Waseda University

In 1886 the Romajikai published a romaji system. Also in 1886 Hepburn published the third edition of his dictionary, which roughly used the system. I've had trouble finding the exact nature of the influence between Hepburn's system and the Romajikai, but they were on friendly terms and cooperating. Hepburn's dictionary's third edition took off, the romaji system mainly came to be called by his name, and it became widely established, particularly among foreigners dealing with Japan.

If you're interested in learning more about Hepburn's dictionaries, a wealth of information is available via Meiji Gakuin University Library, including a comparison of romaji in all published versions and manuscripts.

The Origins of the Nipponsiki System

Around the time the Romajikai was established, one of the members, Aikitu Tanakadate, developed his own romaji system, based on some existing informal conventions, that was specifically designed to mirror the structure of Japanese kana without regard to foreign pronunciation.

In 1885 Tanakadate proposed this new system to the Romajikai, but they ultimately rejected his proposal in favor of keeping Hepburn. As a result Tanakadate left the Romajikai and promoted Nipponsiki on his own. Before becoming known as "Nipponsiki" this style of Romanization was known as "Tanakadatesiki".

The disagreement between Tanakadate was only the first in a long series of arguments between Hepburn and Nipponsiki proponents.

Tanakadate was around 30 when he invented this system, and most of his later career was in physics. He worked with Lord Kelvin in Scotland and helped bring about the adoption of the metric system in Japan. He used romaji in personal writings and promoted it throughout his life. We'll be seeing him again later.

Standard Form and the Romaji Hirome Kai

In 1905 the ローマ字ひろめ会 (Romaji Hirome Kai, "Society for the Advancement of Romaji") was established. They published a magazine, called simply "Rômaji", to create more reading material in romaji and spread its use within Japan. By this point Toyama and Yatabe had both passed away, and it's unclear if the original Romajikai still existed in any capacity. An important early member of the new society was Katsuji Fujioka, who wrote the 羅馬字手引 (Romaji Tebiki), a kind of romaji dictionary.

Shortly after it was founded the Romaji Hirome Kai proposed a slightly modified Hepburn and called it 標準式, or "Standard Form". Note that this is "Standard Form" in general, not "Standard Hepburn", suggesting this was some kind of official standard romanization system; this may have been a response to Tanakadatesiki becoming known as "Nipponsiki", which could imply that other romaji systems were "foreign". As we'll see in a bit, this rhetorical trick ultimately didn't help official adoption.

I haven't been able to track down the original document with Standard Form, but the Romaji Tebiki was already using it in 1906.

Standard Form replaced macrons with circumflexes and always romanized ん as "n", rather than sometimes using "m" depending on the following consonant.

Tanakadate was still around and joined the Romaji Hirome Kai at some point, but got fed up with their use of Hepburn, left, and founded his own organization in 1914.

The Origins of Kunreisiki

In 1930 the government formed a committee to evaluate various romaji systems for official government use. At this point Tanakadate was a member of the House of Peers, and Romaji Hirome Kai member and Hepburn proponent Fujioka was not a politician, but had political influence due to his long and prominent career as a linguist. Both were involved with the government committee.

While the system used at this point varied considerably by government agency, Hepburn was more popular, so it seemed slated for official adoption. However, in 1935 Fujioka died of illness. In 1937, following a recommendation from the committee, a slightly modified version of Tanakadate's Nipponsiki, dubbed Kunreisiki, was established as the official government romaji system.

The main difference between Kunreisiki and Nipponsiki is that kana that are not pronounced differently in standard modern Japanese are spelled as their more common counterpart (づ and ず are both zu), though there are other differences, like how particles are represented, and how ambiguous phonetic boundaries are written.

The War

Despite the recent official adoption of a government-approved romaji system, around 1938 a number of more radical romaji proponents were arrested as potential (or actual) anti-nationalists. One example was Esperantist and language researcher Hidekatsu Saitou, who contracted tuberculosis in prison and died in 1940.

Following World War 2, the occupying American forces rescinded the law making Kunreisiki official and ordered the use of Hepburn for public signage and documents. During the occupation, an American educational commission recommended native Japanese writing be replaced with romaji. This never happened, though romaji became a standard component of post-war education.

Post-war the development of romaji became less dramatic, as serious proponents of replacing Japanese writing systems with romaji mostly disappeared. There have been a number of laws and rules issued by the government over the years, as well as a few special systems like the unique system developed for a textbook called "Japanese: The Spoken Language", but Hepburn and Nipponsiki/Kunreisiki remain the major systems.

There is also much to be written about interaction of romaji with computers, but that's best left for another time.

What About Revised Hepburn?

Sometimes you'll see references to "Revised Hepburn", but nobody knows what this is. It is used to mean, variously:

  • Hepburn's system in his third edition, in contrast to earlier editions
  • the 1905 version by the ローマ字ひろめ会 (properly 標準式 or "standard form")
  • a version introduced with Kenkyuusha's "New Japanese-English Dictionary" in 1954, subsequently adopted by the American Library of Congress
  • other minor variants proposed at different times

While the Library of Congress is surely consistent in their romaji usage, their choice of a particular 1954 dictionary is a bit odd, and distinguishing it from the similar 1905 system is a source of confusion. At one point it seemed like Hepburn might become an ANSI standard, but instead Kunreisiki was adopted, and that standard (Z39.11-1972) has since been deprecated. This lack of an official standard contributes to inconsistency in what "Revised Hepburn" means.

In contrast, "traditional Hepburn" is somewhat reliably used to refer to the system used in Hepburn's third edition as opposed to later forms.

Another common variant, "passport Hepburn" refers to the system previously mandated for names on Japanese passports, which included some exceptions to normal Hepburn, but those requirements are no longer strictly in force, and you can use any romanization you want (such as "George" for what would be "Jouji" in Hepburn) if you have documentation to support it. In practice the most notable feature of "passport Hepburn" was allowing long o-vowels to be written as "oh" ("Satoh"), a practice originating in sports jerseys for some reason. (I have seen mention that the passport rules were relaxed in 2005 but am unable to find any official document concerning this.)

Cutlet Can Help

Now that you know where romaji came from, you can use cutlet to generate it. cutlet is a library and command line application in Python that I wrote primarily to ease the use of romaji in situations where it's required for technical reasons.


I gathered this history to figure out where all the systems had come from and how they came about, in an attempt to clear up a lot of confusion I see in documentation on romaji. What surprised me the most about the history is that while practical considerations have played a role over time, continued effort by a small number of people over a wide span of years is the principle determiner of the systems that still surround us today. Ψ