Use of macron or circumflex in romanzation of Japanese. Fonts with diacritics available commercially or on web. Creating fonts with macrons. Other Macintosh / Windows solutions. Monbusho (Japanese Ministery of Education) and romanization. Search/replace for circumflexes or macrons on Microsoft Word 98 (J).


Archive of messages exchanged on the PMJS mailing list between May 2 -9, 2000. 36 messages.

Question raised by: Alexander R. Bay
Discussants: Anthony J. Bryant, Liza Dalby, Tom Hare, Monica Bethe, Peter Hendriks, Carole Cavanaugh, Richard Bowring, William Londo, Mikael S. Adolphson, Michael Watson, Todd Brown, William Bodiford, Roberta Strippoli, Amanda Stinchecum, Amy V. Heinrich, Philip C. Brown, Michael J. Smitka, Wayne Lammers, Bjarke Frellesvig, Kate Wildman Nakai, David Pollack, Alison Tokita, Rein Raud, Matthew Stavros, Morgan Pitelka.

Signatures omitted or abbreviated. For who's who see

Archive for the PMJS mailing list. Copyright of each message belongs to its author. See general note on editing.

return to: PMJS archive index / PMJS index / members / translations

From Alexander R. Bay

I have a technical question that hopefully someone in the group can answer. I have a Mac, and am using Word 98. I want to get my computer to make the diacritical marks above the O's and the U's but can't figure out how to program my word program to do it. Is anyone out there using Mac, and if so, can you please tell me how I can make the long O's and U's in romanizing my Japanese.

Alex Bay
PhD Student
Pre-1600 Japanese History

From Anthony J. Bryant

It depends on the diacritic.

ASCII code lacks macron vowels, so you will have to use something like the circumflex.

Type "option-i" and then the vowel, and the vowel will have a circumflex.

Option-e gives acute accents.

Option-tilde gives grave accents.

Option-u gives umlauts.

From Liza Dalby

Regarding Alexander Bay's question about how to make macrons for ong "u" and "o":

The question is getting a font which has these marks. What I did was modify the circumflex marking for a particular font by using the program Fontographer. That way, whenever you type the circumflex character it will appear in your printout as a real macron. However, this doesn't translate electronically unless the other person also has that modified font--and will appear as a circumflex. For the text I've written for my website, I just use the circumflex as a substitute macron. At least it's better than doubling the vowel. . .

From Tom Hare

It may be that you're looking to put a real macron over your o's and u's, and that does take some trouble. The Semitica font has the capability, and you enter the character by typing o or u and then typing option-m. There can be complications, though, when you want to put a macron over an uppercase O or U.

The reason I address this reply to the whole pmjs group rather than to you alone is that I'd like to hear what people think about this problem. I went to lots of trouble years ago trying to insert macrons into text while using MS-DOS. ( How distant and arcane that seems now!)

My question, though, comes down to, why do we continute to use macrons at all? Apart from acknowledging a practically meaningless historical connection of our disciplines to Classics, where they were used to indicate long vowels in Latin and Greek, what is the significance of macrons? I cannot think of any reason that macrons are superior to circumflexes, whether in typed or word-processed, or even typeset text. There's no ambiguity to using circumflexes in romanized Japanese and they're far easier and more widely available in electronic fonts than macrons. Why don't we just give up the macron habit (and be thankful it's not Sanskrit we're romanizing)?

From Monica Bethe

On a Mac it is easy to make European diacritical marks in any English font (not Osaka, Heisei mincho, etc.) . Generally the the French circumflex (roof) is acceptable as a substitute for a straight long mark. For the roof press option and "i" and then the letter (o or u), for umlaut (two dots) press option and "u" and then the letter. For other accents, the same system works, generally on the appropriate letter. Check it by opening your key board under the Apple ["Key Caps" in Apple Menu] and pressing option.

If you wish to make a straight long mark, you must create the letters and make a special font. I have two: one in Palantino, one in New York. If you are interested in these, write me. Their problem is that they are font-restricted and the long marks read differently in other fonts, or on other people's computers.

From Peter Hendriks

There are various ways to do macrons, and there are apparently also font
sets available for downloading that contain them. (See H-JAPAN, below)

As far as Word is concerned, I have not used it since it was 5.1a, but assuming that these things have not changed so much since then, it is possible to do macrons in Word using the mathematical formula commands. For a 'o' with a macron do the following.

1. Make sure that your para markers are showing (if not, type COMMAND+J)
2. Enter the formula character by typing COMMAND+OPTION+backslash
3. Type 'o' without the ''. The 'o' stands for 'overstrike'.
5. Type COMMAND+J and this will show you what it looks like when it gets printed out.

For a 'u' with a macron, follow the above steps but in step 4 type (u,SHIFT-OPTION-COMMA)

This issue has come up often on H-JAPAN. See the following address and follow the links to the H-JAPAN archives.

From Carole Cavanaugh

Circumflexes, yes!!

Thanks, Tom.

From Richard Bowring

Re. macrons. Tom Hare's radical proposal does have considerable merit, although it would really have to be a field-wide decision. I, like Liza Dolby, have never had any problems with macrons since using Fontographer about ten years ago to create my own personal font, which incidentally includes lots of diacritics for Korean and Sanskrit as well. The advantage of using this programme is that you can create excellent looking italic and bold versions (although it takes a little time). The only difficulty that emerges is when you send text electronically to someone else who does not have your font. I suppose we could collectively agree on a font created in this way and then distribute it across the network, but then one would get into copyright problems with a) Fontographer and b) Adobe. Any other ideas?

From William Londo

I have communicated privately with Alex about this diacrit issue, not being sure that this question would be of general interest. There is a font, called hobogirin, developed for use by the Hobogirin Institute and the NCC Center for the Study of Japanese Religions, Kyoto, that has the macrons built in. I can provide this font via e-mail to anyone who would like it. It's especially useful to anyone who might want to publish in Japanese Religions, the journal of the Study Center. As a former copy editor, I hope I can discourage the use of circumflexes, though as Liza Dalby points out, it beats kooky romanizations. There's another font, called appeal, available from the International Institute for Zen Research at Hanazono University, that also provides macrons as well as many of the sanskrit macrons. For people who are interested in fonts that will provide macrons, etc., on the PC, I can help with that, too.

From Mikael S. Adolphson

Professor Hare makes a very valid point. Although in principle I prefer (though it might be meaningless, as Hare points out) to use the proper diacritical marks (especially in published works), there are in fact quite a few problems with our dear macrons. Most important, there is a lack of standardization (which in part can be blamed on the software industry's lack of attention to scholarly needs and customs). We use a variety of ways to create macrons; glossaries created in Word 5, the semitica version, Microsoft's recently released macron fonts for Times (available at their web site).

These "solutions" naturally create problems in the exchange of manuscripts and texts between computers and users. The most serious consequence might be the submission of manuscripts to publishers, which, unless they have the exact same fonts (including them with the disks and the manuscript does not guarantee anything), will result in a large number of corrections at the page proof stage. In turn, this may lead to grave errors in the published book itself, which could have been avoided by using a standardized format.

At this point, the standardized format, and the best way of avoiding possible errors at the publication stage, is to do exactly what Professor Hare suggests --- use thosê ûgly circumflexes! At least until a standardized form of macrons appears.

From Michael Watson

It would be useful to pool information about macrons. I'll put together a web page with links to places where fonts with macrons can be downloaded. If anyone has a freeware font that could legally be posted on the pmjs site, I'd be happy to do so.

As a Mac user since 1989, I've always produced macrons for printing with the font "Romance" purchased from

Like the Semitica font described by Tom Hare, macrons are produced by typing o or u and then typing option-m. No problem with capitals here: type O or U and then type option-shift n. Many TrueType fonts from Linguist's Software contain macrons, as well as every conceivable diacritics. Good documentation.

If you _just_ want macrons and don't want to pay for the font, there are some available online. Some Pacific languages need macrons, Hawaiian for example (free fonts Windows & Mac--untested by me):

For other links, see

Liza Dalby mentioned the trick of using Fontographer. Here is good page describing how to create macrons by modifying an existing font:

The problem with this trick is (1) it will display and print fine on your own computer but not elsewhere, and (2) you cannot legally give the modified font to someone else, as the rights to the font still belong to the company that produced the original font.

I assume that this is why the historian Mark Ravina does not permit open downloading of the font with macrons he developed for his students in his "Japan to 1750 class" (how appropriate)--access is restricted to Emory University. But perhaps Alex Bay should see whether Professor Ravina can be prevailed on to share it with a fellow medieval historian. (It would be nice if he would join pmjs too!) See character chart at:

If anyone has recommendations for fonts with macrons for Mac or Windows, please write to me off-list.

From Todd Brown

With the much earlier version of Word that I use, there is a way to create macrons in certain common fonts (like Time) by using something called "overwriting"; though the procedure involves several steps, it is not especially difficult, especially if one first uses circumflexes and then uses "find & replace" to change them to macrons. It might be worth checking the manual to see whether overwriting is a possibility with Word 98.

A better solution, however, would probably be to download some of the free fonts available from Frank Hoffman's Korean Studies web page. These include diacritical marks for both Japanese and Korean, and are very easy to use. Here's the URL:

From William Bodiford

At 5/3/00, Richard Bowring wrote:

Re. macrons. . . . .
<snip> . . . .I suppose we could collectively agree on a font created in this way and then distribute it across the network, but then one would get into copyright problems with a) Fontographer and b) Adobe. Any other ideas?
Richard Bowring

Why not just use Unicode fonts?

With Unicode not only do you have access to every type of glyph and accent one could ever desire, but ones files can be read by any and every other machine that has Unicode fonts installed. It already is the "collectively agreed upon font."

MS-Windows '98 and 2000 already support Unicode. Macintosh System X will support Unicode in the near future.

Right now I am using TITUS Bitstream Unicode Font without difficulty and much delight on my Dell laptop under Windows '98 and MS-Office. The version I have supports about 20 code pages, including extended Latin-A & B (i.e., diacritics galore), IPA, Greek, Coptic, Hebrew, Devanagari, Georgian, Arabic, and countless ligatures and symbols ---- all in a single font! It is 2067k large and contains a total of 6,391 glyphs. Future versions will contain CJK as well. The entire Unicode standard supports around 20,000 glyphs.

The TITUS Bitstream Unicode Font is distributed free by Jost Gippert (, a scholar of linguistics who specializes in Indian and Tocharian languages. These fonts are a beta version of what will eventually be a much larger set, including various Indian and Central Asian scripts. They were developed through the generous help and cooperation of Bitstream Corp., which is allowing the fonts to be freely distributed without a license. For more information, see: TITUS (Thesaurus Indogermanischer Text-und Sprachmaterialien) project at Frankfurt University (

From Roberta Strippoli

Hobogirin and other privately created/modified fonts may do the trick, but as Prof. Bowring said the problem remains: what happens when you give a document written in hobogirin to someone who doesn't have this font? You will always have to send the font along with the document, which for many people is a big hassle.

The best thing in my opinion is to do the jump and liberate circumflexes from their European languages boundaries. I agree with Tom saying we all know we are romanizing Japanese, and we all know what that line/roof/dot/whatever-it-is means in a Japanese language context.

If you still don't feel like doing it... uhm, the next best thing could be agreeing on a common font that does not look too crazy when accidentally read by a computer that does not have it. For example, a font based on --let's say-- Palatino, but where the circumflexes have been genetically altered into macrons. Every time I type a circumflex, a macron comes up. If a computer that does not have the altered font reads the document, it will think it's Palatino and read the macrons as circumflexes, so the damage is reduced to a minimum. Does it make any sense? The problem is that I have no idea how to make such a thing.

The third best thing could be this Hobogirin (which I would be very happy to try) or a font I happen to have, called Eurotimes (don't know if it is copyrighted). It looks exactly like Times, but can do all the macrons, uppercase O and U included. If a computer does not have it, it will read it as Times. The macrons won't work, but at least the document will look very similar to the original. If anybody is interested, get in touch with me privately. I am curious to know where this Eurotimes comes from. I got it from one of my professors a long time ago. Was it created by any of you?

From Amanda Stinchecum

Circumflexes yes!

(I have a PC not Mac. Nevertheless, the question pertains).

From Amy V. Heinrich

Current versions of Word for PCs have a relatively simple way of using macrons, for which you can make up, quite simply, your own shortcut key (I use a two-stroke combination) -- from the "Insert" menu, you can pick "symbol," and use the "multinational extensions" chart, which has both lower case and upper case macrons (even over i and e for gairaigo etc). It will prompt you on how to make a shortcut, which is to use the keystroke combination (<alt -> works for me) and then say, apply.

I use it with Times New Roman. It's not as good as it could be by any means, and it took MS a long time to add the multinational code, but it is really simple and for everything short of camera-ready copy, good enough.

From Philip C. Brown

Roberta Strippoli wrote:

The best thing in my opinion is to do the jump and liberate circumflexes from their European languages boundaries. I agree with Tom saying we all know we are romanizing Japanese, and we all know what that line/roof/dot/whatever-it-is means in a Japanese language context. Cheers,

Ultimately, the expanding adoption of Unicode is solving this problem. I wonder if editors at the Journal of Japanese Studies, The Journal of Asian Studies, or other professional journals in Europe, Australia, etc. would be willing to accept substitutes for macrons. Manuscripts submitted to the JAS and JJS will be returned with requests to pencil in macrons -- their typesetters will not do a search and replace. In addition to these "masters of the universe," my recollection is that macrons have a standardized function in the International Phonetic Alphabet (long vowels).

Unicode is not perfect as a solution (problems with printer drivers not printing what you can display on a screen are still something of an issue), but it is a foundation. Rather than starting from ground zero and creating a replacement for currently available solutions strikes me as unproductive and inefficient. Communication with software manufacturers (including drivers) encouraging uniform adoption of Unicode standards will, IMHO, do much more to solve the problem in a satisfactory way and in a way that will facilitate cross-platform communication.

From William Londo

It seems typography is something we do need to talk about... I'd like to weigh in on some of the messages that have been submitted, as someone who has both written for scholarly works and worked as a copy editor preparing texts in electronic form for publication in a journal. First, what I call "doubling over," of the type suggested by Todd Brown and others, is a really bad idea if you need to send a text to anyone else for publication (or even just printing). This is the procedure whereby you overwrite a character with the diacrit you want. Anyone receiving that text must match your system almost exactly if they hope to read it correctly. Worse, in the case of an article submitted for publication, doing so puts special codes into a text that are very difficult for a copy editor to chase down and convert to the form used by the publication. Making the copy editor do this takes him or her away from more productive things he or she could be doing to make the text better.

The better choice is to find a font in which characters with the diacrits you need are built into the font's character set. Hobogirin and Appeal are two of these for the Mac, and there are others that other people have mentioned, including ones used for Hawaiian, central European languages, etc. For windows, to name just one example, Jim Heisig at Nanzan Institute has developed the New Baskerville and Optima fonts for Windows that they use in their publications. He seems quite willing to share these, but probably best to contact him directly via and ask for a copy if you want one. They are nice fonts.

In any case, whenever you have an article accepted for publication, ask the editor whether they use Windows or Mac to do their publishing, and then ask them for a copy of their font if their system matches yours. Some have noted the hassle of switching to different fonts, but any font can be installed in an instant, and most word processors' find and replace commands can go through and put the special characters right fairly easily. It's possible publishers will refuse to give out their font for proprietary reasons, but this would be astonishing to me given how much easier it is for the editor to produce a well-edited product when the text is in the correct font from the beginning. Can you leave the font to the copy editor to do? Sure, but putting your article in the correct font yourself gives you more control of how your article will look and also opens the way for everyone involved to focus on what's really important in turning out writing worth reading.

Finally, William Bodiford is right. We should all get on board the unicode standard. The faster we do so, the faster all these problems will be relegated to a dusty news thread. I don't believe much in panaceas, but if there ever was one, unicode is it. Not only does it solve all our diacrit problems, it solves all our kanji problems as well. Michael Watson has already put up the Mojikyo web page on PMJS, and though its documentation is not quite at the level I'd like, it's very valuable for anyone who needs to use Japanese script in their text. My hat is off to Michael for connecting us to Mojikyo, and to Prof. Bodiford for putting us onto the Titus website.

Please forgive the length. Stepping off soapbox and yielding the floor....

From Michael J. Smitka

As per Tom Hare's comment, why do we need macrons at all? If you don't know Japanese, they add little or no value. If you do know Japanese -- and here I must admit this is based on my experience as a user of materials Taisho and later -- they are redundant. No one tries (I hope) to convey large blocks of text in romaji -- though I spent several years in a Japanese bank in pre-fax days, when interbranch communication was all in telegraph-style romaji, so it can be done.... If you really need to communicate the underlying text, then simply insert the character. (With or without macrons, you really need the characters for an unusual name, plus it saves one from the embarrassment of using the "normal" reading for a name whose bearer [or bearer's parents] had an idiosyncratic one. Anyway, Japanese fonts are widely available and the encodings are standardized, unlike fonts for macrons.

I assume the real problem lies with (i) tradition and (ii) editors who apply standards relevant for other languages to Japanese. Or does someone out there really rely upon macron as essential to their research?

From Wayne Lammers

I wrote to Ken Lunde, who is in charge of CJK font development at Adobe, about the macron issue. FWIW, he sent this reply:

Earlier this year, the Adobe-Japan1-4 character collection was finalized. I had been working on this project since 1996. I ended up adding 6,090 CIDs (glyphs) with this Supplement. The good news is that macroned vowels are included, in both Roman and Italic forms. Actually, a complete set of italic forms are included. If anyone would like to see all the glyphs in such a font, Adobe Tech Note #5078 has an entire such font embedded:
Tech Note #5078 is about 6MB, and a PDF file. The Roman macroned vowel range is CIDs 9361 through 9370, and the italics are at CIDs 9681 through 9690. Later this year we plan to release Adobe InDesign-J, and we are working hard in order to be able to bundle with it one Adobe-Japan1-4 OpenType font (I cannot reveal at this time what font it will be, but it will have all 15,444 glyphs). This will give users access to true italic forms and macroned vowels, both of which have been lacking. Also of interest is that Mac OS X will be bundling six Dainippon Screen "Hiragino" fonts in OpenType format, with the Adobe-Japan1-4 glyph set (which again means true italic forms and macroned vowels). This will happen hopefully this Summer. This means macroned vowel access on MacOS. I hope that this is good news, and please share it with other members of your list. Regards... -- Ken

As a Mac user, I'm pleased by what he says about OS X, but obviously it will still be a considerable amount of time before anyone can routinely assume when exchanging files that their correspondents have the same or similar macron capable fonts.

From Bjarke Frellesvig

I have been following the macron versus circumflex discussion with interest. I personally use the circumflex and find the macron aesthetically displeasing. Other people think the opposite and will go to great lengths to use macrons in their texts. Apparently there is an element of nostalgia involved as well.

What interested me more, though, was the universal and (almost) tacit assumption that (some of) the so-called long vowels be noted by diacritics.This practice reflects very poorly the reality of the language and is, of course, even worse for the transliteration of pre-modern Japanese. It also effectively prevents any indication of accent. The conventional application of this system is furthermore inconsistent (with 'ei' or 'ii').

The sounds in question are not 'long' vowels, but sequences of identical vowels. They should be written as such (making it possible to transliterate Japanese on any typewriter and still distinguish between Mr Ono and Mr Oono). Even Kenkyusha have adopted double vowels in their excellent Learner's Dictionary, presumably out of a wish to reflect the language as accurately as possible for the benefit of learners.

Briefly: what is it about writing double vowels twice that is thought to be so bad?

(Yes, I confess, I am a linguist.)

From Anthony Bryant

Michael J. Smitka wrote:

As per Tom Hare's comment, why do we need macrons at all? If you don't know Japanese, they add little or no value. If you do know Japanese -- and here I must admit this is based on my experience as a user of materials Taisho and later -- they are redundant. [...] Or does someone out there really rely upon macron as essential to their research?

Hell, yes.

There's a difference between the Mori family and the Môri family.

I know someone who collects koto. Or is that kotô?

Why do we need *any* diacritics? Why not write "J'aime passer l'ete sur la cote" instead of "J'aime passer l'é'té sur la côte"? How about Fernao instead of Fernaõ?

We use them because they make clear distinctions. They have a function, a purpose, and they serve it well.

From Michael Watson

Bill Londo is right about the dangers of making macrons out of umlauts (circumflexes, Danish accents... I seen them all). But I'd still like to thank Todd Brown for introducing Frank Hoffmann's site at Harvard. For those who need fonts with macrons now, his freely distributed TrueType fonts are simple to download, install and use. Just don't expect the diacritics to come out correctly when you send the file to anyone else. But then that's true of my trusty Romance too. For the time being, we're all much safer off using circumflexes when exchanging files.

In the long term, however, Unicode will make life much easier, both for macrons and kanji. For further information of what Unicode means now and soon for Windows/Mac users in East Asian studies, I'd recommend reading Frank Hoffmann's clear explanation:

Among the many benefits of Unicode, I look forward to being able to write web pages mixing kanji and diacritics, now something that cannot be done (hence my macron-less list of translations from classical Japanese--I decided it was more useful to include kanji).

However we'll still need fonts like Konjaku Mojikyo and TITUS that extend the Unicode still further.

The question of fonts and the law has mentioned several times already. I was intrigued by Frank Hoffmann's explanation how he stays on the safe side of the law: it is legal to imitate the look of a propriety font as long as the code is different, and the name is not the same (he calls his Time*, NewYork*, etc.).
Different coding rules out fonts modified by Fontographer.

From Kate Wildman Nakai

From an editorial point of view I'd like to encourage the use of circumflexes, at least for the moment. At MN we have our own macron font, created by our printer, which automatically converts circumflexes to macrons (it's a problem sometimes for true circumflexes!), but as has been pointed out by several people, macrons created in one specialized system or another may look nice in your own hardcopy printout, but they can be a real headache when converting a file to another system (and the problem is compounded if the file also includes kanji). Prospective authors might keep in mind that whenever a submission is accepted one of the first things we ask the author to do is change macrons to circumflexes in the file sent us for copyediting.

On a related matter, another copyediting headache is kanji produced in one of the kanji kits appended to an English-language software system. I'd appreciate any advice that people knowledgeable about such matters might be able to offer. We use Macintosh (presently Word 98--Japanese system) for copyediting and QuarkXpress (again Japanese system) for layout.

From David Pollack

Those working in premodern Japanese can always avoid the need for diacritics in romanization by simply transliterating the "real " orthography, desefu? (or tesefu for the hyperdiacritical purist.) The slight problem is that the general public might not immediately recognize "haseo" as the name of the famous poet.

From Tom Hare

A minor note: When I asked why we need macrons at all, it was not because I think we should dispense with indicating the distinction between "short" vowels and "long" vowels in Japanese. It is often important to make that distinction.

My point is rather a practical one. A central aim of the phonetic representation of language in transliteration is clarity. A circumflex will accomplish that as effectively as a macron, and since it is far easier to produce given our current writing technologies, why not just do it that way?

I don't have anything against aesthetics, quite the contrary in fact, but for my part, the aesthetic distinction between a macron and a circumflex seems slight, and not unequivocally favorable to the former.

Let's hear more from Bjarke Frellesvig, or any other linguists who might be reading the list, about his contention that those vowels for which we need macrons or circumflexes are not "long vowels" but "double vowels." Maybe the distinction is a small one, but I'm not sure I can accept his contention. It seems to me, for instance, that the vowel in "today" (kyo- or kyoo or kyou) is as likely a "long" vowel as a double one. There's nothing, for instance, in the pronunciation of the word which would lead us to suspect two o's, no glottal stop say, and the orthography in kana, though it writes "o," then "u," does not reflect pronunciation or provide us other reasons to characterize the vowel as double rather than long. Or am I missing your point, Bjarke?

From Bjarke Frellesvig

The clearest arguments for the bipartite status of the 'long vowels' come from accent. Whereas in Tokyo J only the first half of a 'long vowel' can carry the accent, in other dialects such as Kyoto J both halfs of the 'long vowels' can do so.

The general view, I think, is that the nucleus, or to some, rime, of long syllables, i.e. syllables with 'long vowels' and syllables ending in /N/ or /Q/ ('small tsu'), consists of two parts, a main vowel and a second element, which can be a consonant, a vocalic element which has some phonetic contrast with the main vowel (an example of such a syllable is /kai/, i.e. syllables with diphthongs), or a vocalic element which has no phonetic contrast with the main vowel; these last are the syllables with 'long vowels'. Strictly speaking, it may be a simplification to speak of double vowels in such cases, but for purposes of transliteration it is a justifiable simplification. Linguistic purists might prefer to write /toRkyoR/ or /toHkyoH/ instead of /tookyoo/, with /R/ or /H/ standing for a vocalic, noncontrastive second half of a long syllable nucleus (or, syllable rime). That is, of course, off-putting in the extreme for purposes of transliteration. Hence double vowels.

I can recommend Timothy J. Vance, 1987, An Introduction to Japanese Phonology, SUNY Press, but decent grammars of Japanese have a section on the phonology explaining these things one way or the other.

From Monica Bethe

Has anyone gotten wind of the new Mombusho dictate (maybe this is just a rumor, I heard it on the radio) that long "o" be written "oh" instead of with a macron??? Kyoto will, according to this, be Kyohto and Tokyo, Tohkyoh. No mention was made of what they
intended to do with the long "u". Maybe our entire debate on macron will prove to be of no avail......

From Alison Tokita

Everyone must have noticed that the Japanese are terribly inconsistent in romanizing, except perhaps those purists who self-consciously insist on kunreishiki (sorry, kunreisiki) as taught to Grade Four students. (This turns Mt. Fuji into Mt. Huzi and so on. Of course, to be quite consistent to Japanese orthography, it should use not d but t", not b but h"...) This system requires the use of the circumflex, I think. On the whole though, the Japanese seem to want to ignore the long (or repeated) vowels when romanizing, and most write their names for example as if they were short, e.g. Ono, Oki, Okura etc. Some will however use oh in such names. (Many of us researching the performing arts also do this with noh drama.) In transferring hiragana to romaji (yes, I know it should have a macron) my Japanese students when they write in English will spell Monbushou, or Monbusho, Keisou Shobou or Shobo, not Shoboo, and never use the macron. They don't seem to know what it is.

With the experience of romaji input to type kanji on a word processor, which requires the Japanese orthography, many non-Japanese are also romanizing in this way these days. There are no macrons on the keyboard, and only kyou not kyoo will produce *1. Interestingly, word processing input will take a few variants in romanization (chi and ti both produce *2), and some which are peculiar to word processing, such as thi for *3 and li for small  *4.

 *1  *2  *3  *4

(Don't you remember feeling frustrated before you found this out?)

This leads me to ask: do we as foreign researchers of things Japanese have the right to determine the romanization system. The Chinese have imposed pinyin over Wade-Giles. Will the Japanese be able to impose the Kunreishiki over the Hepburn (Hebon) system? What sort of cultural colonialism are we practising?

Personally, I am wedded to the use of the macron, but have a slight envy for those people who always ignore them. In politics, economics, sociology and other fields, it seems to be the norm not to indicate the long/extended/doubled vowels. Sometimes I take this course of least resistance if publishing in a medium which can't really cope with the complication, or doesn't want to cope with it. It is tempting to give up the effort, after having produced a whole book with macrons painstakingly inserted, had them fall off when converted to Mac, paid someone to put them in again, and have them fall off when converted to another program for final publication!

Why is it soo difficult to produce the little stroke over the vowel?! Is it worth it?

There are the other problems of romanization, such as when to hyphenate, when to have a break between 'words', when to capitalize. For example, do we write: Muraki-san, Muraki san, Muraki San? Also, when to italicize Japanese words in an English text: every time a word appears, or only the first time? And so on, the problems are legion, and as scholars it would be nice if we had an agreed system which is logical, easy to apply consistently, and true to the spirit of the original Japanese.

From Anthony J. Bryant

Monica Bethe wrote:

Has anyone gotten wind of the new Mombusho dictate (maybe this is just a rumor, I heard it on the radio) that long "o" be written "oh" instead of with a macron??? Kyoto will, according to this, be Kyohto and Tokyo, Tohkyoh. No mention was made of what they intended to do with the long "u". Maybe our entire debate on macron will prove to be of no avail......

I hope to God that's not true.

If that happens... is "dohyo" to be understood as "doh-yo" or "do-hyo"?

Don't they even stop to *think* about these things?

Alison Tokita wrote:

This leads me to ask: do we as foreign researchers of things Japanese have the right to determine the romanization system. The Chinese have imposed pinyin over Wade-Giles. Will the Japanese be able to impose the Kunreishiki over the Hepburn (Hebon) system? What sort of cultural colonialism are we practising?

I don't think we're imposing *anything* on anyone. The use of Romaji is for *us* to enable us to pronounce Japanese. Romaji exists to make life easy for Americans, for French, for Germans, for Italians... I don't know what Russians do when the transcribe Japanese, but I doubt they see it as cultural imperialism imposed on the Japanese. They know what they're doing is to enable a Russian to see a word and sound it out so it comes out like passable Japanese.

If I see huzi, I'm *not* going to think about that mountain. And if I see syotoku taisi I'm not thinking about some regent. I don't know what that is supposed to be.

The problem with kunrei and pinyin is that they were created by committees of government appointees rather than foreign language people who will have to *use* the systems. The success or failure of any system of Romanization lies in the ability of one *not* familiar with any idiosyncrasies of the system to pronounce the words approximately correct at first sight. On the whole, Hepburn succeeds. Wade-Giles, Pinyin, and Kunrei fail miserably. As one who is now taking Chinese, I have to say that I abhor Pinyin with a hatred that burns hotter than the fire of a thousand suns. There is no consistency with vowel or consonant notation, and Wade-Giles is no better.

There are the other problems of romanization, such as when to hyphenate, when to have a break between 'words', when to capitalize. For example, do we write: Muraki-san, Muraki san, Muraki San?

That's a real problem. I see the same people who write "Muraki-san" and "Muraki Kacho^". So which is it? <G>

Also, when to italicize Japanese words in an English text: every time a word appears, or only the first time?

At least this isn't unique to us. That is a common problem anytime one writes about things in a foreign milieu, be it Japanese, French, German, or Russian.

From Alison Tokita

In reply to Tony Bryant

I sympathize with what you say, but I wouldn't claim that Hepburn helps French, Germans and Italians etc. The French for example would probably prefer to romanize [Fuji] as foudji, and [shamisen] as chamisen, because obviously Hepburn follows the English phonology not French. This is true for all other European languages too. So they must use a system which developed for users of English.

From Rein Raud

There are some languages (Czech, Hungarian, Latvian, Lithuanian) that use, at least in part, an indigenous system of romanisation, which makes Japanese words more adoptable to the speakers/readers of these languages as long as they know no other, but creates difficulties for those who take a broader interest in Japan and venture beyond the texts available in their own languages. The Russians (as well as Ukrainians, Bulgarians, Serbs etc.) have the same problem, but since they cannot use any Latin-based romanisation system due to the Cyrillic alphabet there is nothing that could be done about it. Thus it is the question of a more "closed" and a more "open" interpreting community. In Finland, too, some people advocate the use of double vowels instead of macrons/circumflexes, because it corresponds to the normal practice of the Finnish alphabet. In Estonia, however, where the orthographical practice is the same, nothing beside Hepburn has been proposed, although its consonant values do not coincide with the values of these letters in Estonian. Nevertheless, things Japanese do not make up a too prominent part of the Estonian cultural scene, which is why I am strongly in favour of the use of Hepburn here - it includes the Estonian reader in a more open interpreting community and enables to access materials published in other languages as well. In a truly indigenous Estonian transcription, Fuji would appear as Vutsi. A specialist could probably use three or four encoding systems simultaneously for the use inside/outside one's own culture, but e g looking at bibliographies of research published in Czech or Latvian one has to know the basics of the orhtography of these languages in order to determine with any certainty who or what is being talked about.

From Matthew Stavros

Let's not take the Monbusho too seriously now. After all, it has been recently decided that the mathematical ration of the circumference of a circle to its diameter (pi) will be "officially" taught in Japanese schools as being three (3) !!! Can you believe it? Furthermore, the forward-thinking geniuses at the Monbusho (should I put a macron on that last 'o' ?) have declared that computer and internet education will be incorportated into standard school curriculum by the year 2003!! In one of, if not the most, technolocially advanced country in the world, is that revolutionary or what?

Excuse my sarcasm but I refuse to write Kyohto!

From Morgan Pitelka

On the topic of macrons, I have a question for users of the Japanese version of Word 98 (for Mac). This is rather technical, so non Word 98J users may want to skip this message.

I am unable to search and replace phrases with circumflexes or macrons, which is extremely tiresome. For example, let's say that I have a document that contains numerous long "o" characters that I have represented with a circumflex. I want to find all the circumflexed "o" characters and replace them with "o*" so that I can save it as a text file and send it to someone with an old computer and old fonts.

When I bring up the search and replace window, I am unable to enter characters outside of the system font of Osaka (and the roman set for Japanese fonts don't, of course, include circumflexes or macrons). Even if I highlight the text box in the search and replace window and change the font to something with macrons or circumflexes, I cannot enter the circumflexed "o." I have also tried cutting and pasting.

I have this problem with _any_ character outside of the standard Osaka character set. One way to get around it is just to past the relevant text into Nisus, do the search and replace, and then return it to Word. But I would prefer to do the whole operation in Word.

Any advice?

From Michael Watson

As Microsoft Word is widely used, I will send my answer to Morgan Pitelka's question to the list.

Microsoft Word 98 (J) for Mac does not handle well search/replace with circumflexes or macrons, but it can be done.

(1) circumflex o (or macron o) to o* You cannot directly type ô into the top window (kensaku suru moji retsu), but if you copy and paste circumflex o, ^u244 appears instead of the diacritic. Click "subete okikae" (replace all) as usual.

(2) o* (or whatever) to circumflex o (or macron o) Again you cannot type a diacritic in the lower window, and Word refuses to substitute ^u244 . The work-around is this: copy the desired letter with circumflex or macron, but leave EMPTY the lower window (okikae-go no mojiretsu). Pull down "tokushu moji" menu and choose "kurippubôdo no naiyô" (contents of clipboard). The accented character in the clipboard will be substituted for o*.

Circumflex o to macron o can be performed by combining (1) and (2): searching for ^u244 and replacing it with macron o (copied to clipboard).

On Microsoft 2000 for Windows (J) things work a bit better. A diacritic can be copied directly into the upper window. Macron o (a "tokushu moji") would not display directly in the lower window, and there is a warning beep. However macron o can be seen by pulling down on the window (arrow on right), and search/replace works otherwise.

Editor's note (2000/05/12): Converting these messages to web page format posed problems not unrelated to those under discussion. Some writers used diacritics, while one used kana and kanji. Without using Unicode encoding, both cannot be displayed in the same web page. Diacritics were at issue here, so instead of using Japanese encoding (Shift-JIS) as elsewhere on this site, I have used Western character set (iso-8859-1). I broke another rule by using image files to illustrate the kana/kanji mentioned by Alison Tokita in her first mail ("Fuji" and "shamisen" were originally in Japanese script in her second mail.)

return to: PMJS archive index / top PMJS page / members / translations / pmjs mailing list.