Kwannon or Kannon?

Readable romanization

By William Wetherall

This is a slightly reformated version of an article published in 2005 by the Society of Writers, Editors & Translators. This web version includes color scans of the images that appeared as black-and-white images in the SWET version, as well as hot links to websites mentioned in the article.

A version of this article appeared in
SWET newsletter, Number 107, April 2005, pages 3-13

No system of kana orthography, let alone romanization, has managed to account for the diverse pronunciations of Japanese words. Pronunciations of ages past, notoriously difficult to pin down, may be impossible to render in romaji -- but it is tempting to try. Inspired by an Internet conversation on how to spell the name of a historied bodhisattva, William Wetherall offers a delightful defense of romanizing for readability but also reminds us of reasons to consider variant spellings. Some of Wetherall's articles on other subjects, translations of short stories, and his own short fiction are accessible from

An author who writes historical fiction set in Japan, in English, recently queried the premodern Japanese studies (PMJS) list forum at with an interesting question. She wanted a clarification on the spelling of Kwannon versus Kannon. She had written Kwannon, thinking that "the Chinese Kuan-yin probably accounts for Kwannon, and this might have been more likely to have been preserved in the eleventh and twelfth centuries." Her editor, though, considered Kwannon outdated.

Members of the forum, including myself, contributed commentary on historical pronunciation, orthography, and romanization. Some of us leaned toward Kannon. Others found Kwannon also acceptable. My blunt afterthoughts: (1) both the author and the editor are wrong, (2) either Kwannon or Kannon would do, (3) Quannon might be kooler, (4) Kannnon may be the wave of the future, and (5) Canon is the most photogenic.

KwansyejH[?]im and Kwanzeom

One writes or translates fiction to tell a story, not dwell on linguistics. Most authors who write historical fiction in Japanese today pay no attention to contemporary orthography. So why should an author of a story set in the Heian period, but told in English, worry about whether Avalokitesvara was Kwannon or Kannon? Especially since then, as now, there were many pronunciations, and no one really knows how Kannon was pronounced in China at the time of the Nara or Heian periods.

While Chinese pronunciations surely informed early Sino-Japanese, it does not follow that Kuan-yin accounts for Kwannon. A reconstruction of the Chinese that spawned Sino-Japanese relies on analyses of Chinese, Korean, and Japanese texts, just as a reconstruction of Japanese dialects during the late sixteenth century and the Edo period depends on studies of Japanese, Portuguese, Dutch, English, and other records of Japanese speech.

Middle Chinese, a composite of Chinese dialects spoken at the time Japan was intensely absorbing Chinese civilization, largely through Korea, differentiated between "-n" and "-m." Man'yogana, and the kana that replaced it, did not include "-n," which did not fully develop in Japanese speech until the end of the Heian period.

Kannon ŠΟ‰Ή is an abbreviation of Kanzeon ŠΟ’‰Ή, arguably the more prevalent name in Nara and Heian times, though today ŠΟ’‰Ή will draw only one-tenth as many hits as ŠΟ‰Ή on Google. ŠΟ’‰Ή is the Chinese translation of Avalokitesvara, the genderplastic bodhisattva who "(compassionately) sees/perceives the sounds/cries of the world" -- whence the epithet God/dess of Mercy.

One Middle Chinese reconstruction of ŠΟ’‰Ή is kwan-syejH-[?]im (following Baxter's notation).* ‰Ή was "[?]im"** and only later did it become the "yin" of present-day Mandarin. One standard Japanese translation of Nihon shoki, Japan's first national history, written in Chinese and published in 720, reads ŠΟ’‰Ή as ‚­‚ν‚ρ‚Ί‚¨‚ή (kuwanzeomu > kwanzeom).

* and personal communication from William H. Baxter. See also William H. Baxter, A Handbook of Old Chinese Phonology (Trends in Linguistics: Studies and Monographs), Berlin: Mouton de Gruyter, 1992.

** [?] is a poorman's glottal stop. It is usually written with an IPA (International Phonetic Alphabet) symbol which looks like a question mark without the dot at the bottom (U+0294 in Unicode).

Kwanseum and Kwanum

Nihon shoki (aka Nihongi), based on oral and written accounts, states that Buddhism was introduced to Japan in 552 or thereabouts by Paekche (Kudara), the Korean kingdom with which the Yamato court was then most closely related. By 668, with the help of Tang forces, Silla had defeated Koguryo and Paekche to unite the Korean peninsula and end the Three Kingdoms period, during which the peninsula was dominated by China.

Nihon shoki relates that in 686, Tenmu became so ill he delegated his imperial authority to Jito, his consort and successor. Some officials then made Kanzeon images and recited the Kanzeon sutra for him, but to no avail. Three years later, Silla sent a condolence mission bearing a number of gifts, including a gold-copper image of Kanzeon.

One of Japan's most famous National Treasures is the slim, graceful, two-meter figure called Kudara Kannon (•SΟŠΟ‰Ή) at Horyuji in Nara. The camphorwood statue was reincarnated from storage at the end of the Meiji period. Its name was contrived in the belief that it was brought from Korea during the seventh century. Camphor did not naturally grow on the peninsula, though, so the statue was probably carved in Japan by immigrant Paekche artisans or their Yamatoized descendants.

[Stamp illustration]
[Caption for Kannon stamp]
Kudara Kannon, National Treasure, Horyuji, Nara (issued 1967).

Sino-Korean resulted from several centuries of direct assimilation of Middle Chinese. The Middle Chinese reconstruction of ‹ΰ is "kim," now "jin" in Mandarin. Korean continues to distinguish between "-n" and "-m," and the present-day versions of ŠΟ’‰Ή and ŠΟ‰Ή are Kwanseum and Kwanum.

The lips of the gods

My advice to authors of historical fiction set in early Japan: stick with Kwannon or Kannon. Linguists may get high on this stuff, but other readers will fall asleep.

Trying to squeeze a Rosetta stone out of the sandstone of man'yogana tables, Sino-Japanese readings, Chinese rhyme dictionaries, and Chinese transliterations of non-Chinese words is high linguistic adventure but also very speculative. Readers get a rough ride when authors take historical linguistics too seriously.

The main text of Donald L. Philippi's 1969 (1968) translation of Kojiki is marred by in-your-face romanization, in a distinct font no less, that reflects a hypothetical dialect of the Yamato court of the seventh and eighth centuries. Philippi might as well have written all the proper names in the International Phonetic Alphabet (IPA), as though to suggest he had been there and heard them from the lips of the gods. The letters leap off the page, whereas even the shameful Latin in Basil Hall Chamberlain's 1932 (1882) translation is typographically indistinct.

Philippi shows Hepburn romanizations in a wonderful glossary, but he should have done just the opposite -- render the names inconspicuously in the text, and save the heady linguistic minutiae for the back. The number and color of Yamato vowels and consonants simply have no bearing on the Kojiki stories or their illumination. Philippi might also have adopted more elegant versions of Chamberlain's somewhat awkward translations of meaningful names.

Philippi translation of Kojjiki
showing names in Old Japanese transliterations
(Philippi 1969[1968]:47)

At the time of the beginning of heaven and earth, there came into existence in TAKAMA-NO-PARA a deity named AME-NO-MI-NAKA-NUSI-NO-KAMI; next, TAKA-MI-MUSUBI-NO-KAMI; next, KAMI-MUSUBI-NO-KAMI. These three deities all came into existence as single deities, and their forms were not visible.

Chamberlain translation of Kojiki
showing meaninful names translated into English
(Chamberlain 1932[1882]:17)

The names of the Deities that were born in the Plain of High Heaven when the Heaven and Earth began were the Deity Master-of-the-August-Centre-of-Heaven, next the High-August-Producing-Wondrous Deity, next the Divine-Producing-Wondrous Deity. These three Deities were all Deities born alone, and hid their persons.

Older orthography

Algernon B.F. Mitford, aka Lord Redesdale, like James Curtis Hepburn, was an eye and ear witness of late Edo and early Meiji Japan. His still very popular and readable Tales of Old Japan -- first printed in 1871, fifteen years before Hepburn published his romanization scheme in the 1886 (third) edition of his 1867 dictionary -- is full of spellings like Adzuma, Daijokwan, daimio, Hiogo, Idzu, Idzumi, Iyeyasu, Jiuyemon, Kiyoto (and Kioto), Kwannon, Niurai, Sasawo, Tokio, Uyeno, Yedo, and Yukiye (from 1910 printing of 1874 second edition).

Mitford's spellings were part of the romanization legacy that inspired Hepburn to refine his system. They did not, however, reflect a particular variation of Japanese speech at the time any more than Chinese characters or English spellings today represent a specific dialect of Chinese or English.

A country as vast and geographically complex as Japan cannot help but have numerous dialects. In Mitford's and Hepburn's day, as now, there were many people in Edo/Tokyo, to say nothing of Ryukyu/Okinawa or Ezo/Hokkaido, who would not have spoken the way Japanese was most commonly written. The Heian capital, too, must have been awash with dialects, since nearly 30 percent of the aristocratic clans were of fairly recent immigrant origin, and the relatively indigenous population would also have included provincial migrants.

At best, Mitford's spellings represent transliterations of contemporary kana orthography, now called "historical (old) kana usage" (rekishiteki [kyu] kanazukai). Developed in the late seventeenth century, the orthography differentiated "ye" ("we" ‚ο ƒ‘) from "e" (‚¦ ƒG), and "kwa" (‚­‚ν ‚­‚μ ƒNƒ ƒNƒŽ) from "ka" (‚© ƒJ), among other distinctions that by then were no longer common in Edo/Tokyo speech.

The current kana standard, called "present-day (new) kana usage" (gendai [shin] kanazukai), was promulgated in 1946 after heated debate that began during the Meiji period and continues today. However, the older orthography persisted for several years in postwar publications and can still be seen in letters written by older people. It is also kept alive in some personal and other proper names, and is sometimes used to color things "traditional" -- like ‚Η‚Ί‚€ (dozeu), the Edo/Tokyo version of ‚Η‚ΐ‚β‚€ (dojiyau), in the names and on the menus of restaurants that feature a quick little loach more widely known today as ‚Η‚Ά‚ε‚€ (dojo).

Kana rubi (furigana) in prewar and early postwar newspapers and magazines still followed older orthography, but not as strictly as in late Edo and early Meiji print media, including woodblock prints with stories. Still, if writers or translators in English were to adopt standard Japanese orthography as the basis for romanized spellings, they would not be wrong to write Kwannon or Yedo, or even gwaikokujin, for Japanese up to the middle of the twentieth century -- long after such words had ceased to be pronounced this way by most speakers. The only question, ever, is whether the extra letters are worth the burden they may impose on a reader.

Kwan-ei and Kan'ei

["W-glide on woodblock print" illustration]
[Caption for detail of Kannon woodblock print]
W-glide on woodblock print.
"Kwannon reigen ki" series
(Spiritual experiences of Kwannon).
Kunisada and Hiroshige II (ca 1858-1859).

The w-glide, despite its loss of status in standard orthography, has a life of its own, both inside and outside Japan. Official language reforms and romanization standards reach only so far. In the natural world anything is possible, and numerous pronunciations and spellings that language police and purists would reject as "incorrect" become both proper and authentic.

I am currently translating a story set in late Edo and early Meiji. The author does not bother the reader with older kana for contemporary names. He creates ambience, instead, by dramatizing older styles of social intercourse in older physical settings.

γ–μŠ°‰iŽ› appears in the story. I am writing it Ueno Kan'eiji. Had the author attached furigana to reflect the orthography that prevailed in late Edo and early Meiji, I might write Uyeno Kwan-eiji. While I personally regard such relics as affectations, I would respect the author's attempt to be archaeic or quaint.

Š°‰i is also the name of the Kan'ei period (1624-1644), arguably the two most important decades of the early Tokugawa era. Numismatists the world over would recognize and might even use spelling variations like Kwan-ei/Kwanei and Kanei. They would also recognize the Kwammon variation of the Kanmon-denominated Kan'ei Tsuho copper and iron coins that were minted in huge quantities and used well into the 1860s.

Mitford curiously spells this period Kan-yei, which reflects his tendency to overplay "ye-" and underplay "kwa-" -- possibly because his ear at times affected his eye. In "The Loves of Gompachi and Komurasaki" he refers to the Meguro grave of the inseparable lovebirds as the "Tomb of the Shiyoku." An illustration of the tomb is similarly captioned below the characters ’Λ—ƒ”δ, appropriately written right to left.

Most people today, even in Tokyo, know the tomb as Hiyokuzuka, though true Edokko might say Shiyokuzuka, just as they might say (or even write) Shibiya or shigashiguchi. Shirai (”’ˆδ) hailed from the province of Inaba, now part of Tottori prefecture, where some people insist his name was Hirai (•½ˆδ) and blame Shirai on the playwrights who popularized his legend on Edo stages.

Double standards

The Kunrei system, once a foe, is now a friend -- of sorts -- of w-glides and even Hepburn spellings.

Japan's official romanization standard since the 1937 Cabinet Order (Kunrei) that established it, Kunrei is essentially a modified version of the Nippon system advocated by Tanakadate Aikitsu in 1885, a year ahead of Hepburn. The Nippon system provided for "kwa" and "gwa" because w-glides were part of the standard orthography, but the authors of Kunrei ignored them as dialect variations of "ka" and "ga" in standard speech. What might be "right" in kana would have to be "rite" in romanization.

Then, in 1954, the government issued a Cabinet Order that modified the official romanization standard to reflect the new 1946 kana orthography and other realities. The 1954 order contains a table of alternative spellings that can be used to accommodate international relations and continuing customs (jurai no shukan). The spellings include rival Hepburn forms like "shi" and "ji" but also "kwa" (ƒNƒŽ) and "gwa" (ƒOƒŽ).

If the Japanese government says spellings like "shi" and "kwa" are okay, they've gotta be okay. Or rather the government, in allowing such exceptions, is admitting that it has little or no control over international or customary spelling preferences. The double standard does create not a little confusion, but that's another story.

Proper w-glides

Though the w-glide in "kwa" has disappeared from the standard pronunciation of Sino-Japanese "ka", it remains an easily pronounceable and writable sound, as in names like Michelle Kwan (ƒ~ƒbƒVƒFƒ‹EƒNƒƒ“, less often ƒNƒŽƒ“), whose Chinese name is ŠΦ as in Kwanto (ŠΦ“Œ), Kwansai/Kwansei (ŠΦΌ), and kwankei (ŠΦŒW).

Kwansei Gakuin University (ŠΦΌŠw‰@‘εŠw) in Nishinomiya, aka Kwangaku, formally retains the w-glide in its institutional name. Kansai University (ŠΦΌ‘εŠw) in Osaka does not, but some Kandai teams, clubs, and journals use Kwansai in their romanized names. The art magazine ”Ε‰ζεYp (Hangwa geijutsu), which was launched in 1973, became ”Ε‰ζŒ|p (Hanga geijutsu) from its 101st (September 1998) issue.

Authors and editors cannot dismiss a w-glide in a name without determining whether it is the established formal spelling. In Hearn's Yotsuya Kwaidan it is. In Web references to "Kwansai International Airport" it is not.

Keeper of the keys

The ear is a lover of speech and the eye a whore of authority. Authors may write by Caesar's rules but read according to those of their personal gods. Readers, too, recite written stories according to the colors of their own dialect and idiolect.

This is so because writing is not language. Kanji, kana, hangul, and alphabetic letters represent sounds that can greatly vary between writer and reader. No script encodes or decodes itself, and despite considerable sharing, all writers and readers hold their own phonetic and semantic keys.

Using older spellings to give a medieval flair to a story works only if a reader buys into the illusion that a single letter makes Kwannon looke more mediaeval than Kannon. In any case, why should a romanized spelling invented in the middle of the nineteenth century, to transliterate a late-seventeenth-century kana orthography that was already out-of-date, be more appropriate for a story set in the Heian period? One might as well write Kannon and get on with it.

Kannon can represent and elicit numerous pronunciations. It neither forces one to say "ka" nor prevents one from saying "kha" or "ga" or "kwa" or "khwa" or even "gwa." From the mouths of different readers, the phonemes k, a, n, and o will scatter all over the phonetic chart.

I have assimilated "standard" Japanese and have learned to associate Hepburn romanizations with its pronunciations. Hence I would pronounce ŠΟ‰Ή or ‚©‚ρ‚Μ‚ρ or Kannon as "kannon" (you either share the code or you don't). But how am I to know that the author who wrote these graphic representations would say "kannon"?

I might attend a public reading and hear the author utter "cannon" or "canon," or "kuan-yin" or "guanyin" or whatever. And all this time I've been led to believe it was "kannon"! Who do I sue, the author or my language teachers?

Keyboarding "kwa" and "n"

The editor of the author who wants to write Kwannon apparently tripped over the spelling as out-of-date. However, Kwannon is still found in the names of organizations all over the world, including some in Japan. Though doubtlessly far less popular than Kannon, it has several quite viable variants, each with its own patina.

At around 19:30 Tokyo time on January 14, 2005, Google found 7,700 web pages mentioning Kwannon. It found 161,000 pages with Kannon, and 42 with Kuannon.

In addition, Kannnon got 288 hits and Kwannnon 6. You are and are not seeing triple. These are rapidly becoming the new romanizations, based on how one has to type Kannon in order to produce ŠΟ‰Ή or its kana equivalents on a keyboard programmed for Japanese script. Typing "kanon" will produce kanji or kana for "ka'non," whereas "kannon" will produce "kan'on."

Japanese input editors are programmed to accommodate mixing of Kunrei, Hepburn, and a few non-standard spellings. Typing "cannnon" will produce ŠΟ‰Ή just as surely and quickly as "kannnon." Though "kannon" or "cannon" will eventually lead to ŠΟ‰Ή, typists learn that "nn" is keyboardese for "-n-" (kon-yaku) in old Hepburn and old Kunrei, or "-n'" (kon'yaku) in new Hepburn and new Kunrei, while "nnn" is keyboard-speak for "-nn-" (konnyaku) in both old and new Hepburn and Kunrei.

Microsoft's standard Japanese input editor does not recognize "kwan" or "kuwan" as alternative spellings of the first syllable of ŠΟ‰Ή. In this sense, either Kwannon or Microsoft is out-of-date. However, to the editor who objects to Kwannon in favor of Kannon, I would say butt out. Let the author have her way. Or at least janken. And before coming to blows, consider Quannon.


Google came up with 805 hits for Quannon and 13 for Qwannon. The offices of a group called Amida Trust are located in a building named Quannon House in Newcastle upon Tyne. The UK is also home to the Amida Quannon Buddhist Project, which claims to train Zen therapists.

So is Quannon a New Age mutation of Kwannon? Yes and no.

Names like Quanto are found on a number of early maps. Variations of Quannon, too, are common. Aime Humbert, in Japan in 1863-1864 as an envoy for Switzerland, later published nearly two-hundred woodcut engravings in a volume called Japan and the Japanese (London 1874, French edition 1870). One is captioned "Interior of the temple of quannon." Other captions include "A street in Yeddo on new year's day," "Shops and warehouses at Niphon-Bassi," and "Restaurant, with a view of Fousi-Yama."*

* The National Diet Library features these and other contemporary illustrations in a web exhibition called "Scenic mementos of japan" at and

Meiji, Taisho, and early Showa books in English typically have Kwannon, but Quannon (a Portuguese spelling) becomes more likely as one goes back. Augusta M. Campbell Davidson (Present-Day Japan, 1904) has Kwannon and Yedo. William Elliot Griffis (The Mikado's Empire, 1876), has Kuanon in Asakusa, and Kuanto and Yedo, and the index refers readers from Quanon to Kuanon. Richard Hildreth (Japan and the Japanese, 1860, 1855) has Quanto, Quanwon, and Yedo. In the February 4 (Shonguach [Shogatsu] 4) 1622 entry in his diary, Richard Cocks writes that "we went to see the cytty and a great pagod called Assackxa, dedicated to a Japon saint (or rather deavill) called Quannow." Cocks writes Quanto but has Edo rather than Yedo. The "ngua" in "Shonguach" seems to reflect a w-glide "gwa" in which the voiced velar stop (hard) "g" has become the velar nasal (soft) "ng" (Na-ga-no / Na-nga-no).


Googling "Quannon" also brought up Quannon CAD Systems, a software company in Minnesota. Its products appear to have nothing to do with Buddhism, but you never know. You would not expect religious inspiration behind a camera, either, but you could be wrong.

Google produced another 768 hits on Kwanon, mostly thanks to the legacy of Japan's first 35-millimeter camera, called Kwanon in romaji but ƒJƒ“ƒmƒ“ (kannon) in kana. Kwanon's principle developer and the founder of the company that produced it was a believer in Kannon. A thousand-armed God/dess of Mercy graced the camera's logo, and the namesake of its Kasyapa lens was Mahakasyapa, a disciple of Buddha who devoted his life to spreading faith in Kannon.

Soon after the Kwanon came out in 1934, someone argued that changing the "kw" to "c" would result in a globally more appealing and pronounceable brand name. So in 1935 the camera was renamed Canon or ƒLƒ„ƒmƒ“ (kyanon), which became the name of the company in 1947. The new spelling was linked with "canon" as in "sacred law and model for judgment," which reflected the high-quality aspirations of the company, then called Seimitsu Kogaku Kenkyujo (Precision Optical Instruments Laboratory).

The hundreds of millions of people who say "kwanon" or "canon" for Kannon cannot be wrong. I withdraw my suit with apologies.