Talk:Nihon-shiki romanization

Long vowels

Nihon-shiki uses ō for both おお and おう. Doesn't that contridict the statement that it "allows lossless mapping to and from kana"?

Zeimusu 14:41, 2004 Apr 17 (UTC)

Kunrei allows (and AFAIK even recommends) "Oo" for initial long O, which is the only place you'll find おお; are you sure Nihonshiki doesn't? Jpatokal 09:53, 18 Apr 2004 (UTC)

Actually, 'tis not true. Long o written as oo instead of ou isn't limited just for initials (see below, 通る, for instance). Of course, if oo is allowed instead of ô in this case, it isn't a problem... If not, however, the clause about lossless mapping is simply untrue. How is it, then? I don't know. --213.186.250.141 07:09, 25 October 2006 (UTC)

Original Nihon-shiki [[1]] said nothing about long vowels explicitely. But in example, they use 'Heimen' and 'shikō site'. So "allows lossless mapping to and from kana" is true, but did not allows lossless mapping to and from word or sentence of kana.--RedDragon 07:51, 25 October 2006 (UTC) 　

Kwa and gwa

Relating to this: http://weblog.delacour.net/archives/2003/02/just_make_sure_you_spell_it_incorrectly.php

What about kwa and gwa? They seem to be "outdated" romanizations - How long have they been outdated? Is there any more info on kwa and gwa? WhisperToMe 00:07, 2 Aug 2004 (UTC)

They're obsolete forms of modern "ka" and "ga", reflecting an ancient difference in pronunciation and theoretically written くゎ and ぐゎ, but in practice absolished (AFAIK) in the post-WW2 writing reforms. About the only place you'll run into a "kwa" in English is the equally obsolete spelling "Kwannon" for Kannon (and this only in pre-WW2 sources that also speak of "Yedo" and so on).

The blog you cite is incorrect in that the existence of "kwa/gwa" is still recognized in Kunrei as well. However, since the kana combinations くゎ/ぐゎ are never used in the modern Japanese, there's no need to transliterate them either... Jpatokal 00:24, 2 Aug 2004 (UTC)

Long OH sound

I've had several confusing conversations with native speakers of Japanese over the years, trying to clarify the spelling of words like shoyu (soy sauce), Tokyo, etc.

Are おう and おお used interchangeably, or what? --Uncle Ed 16:28, 17 March 2006 (UTC)

Well, both resolve to long o. Some words are spelled "o-u" and some "o-o". WhisperToMe 17:59, 17 March 2006 (UTC)

They are not interchangeable in Japanese; which is correct depends on the word. おお is generally a kun-yomi (i.e. Japanese) reading, so you have おおきい, とおい, etc. It also appears when you have a character ending in "o" followed by a character beginning in "o," as in the place names ひろお and みのお.

おう is usually an on-yomi (i.e. Chinese-derived) reading, e.g. おうさま (王様), たいおう (対応), although it also shows up as a conjugation of verbs ending in う (e.g. 会おう). This is why you have To-u-kyo-u and Kyo-u-to, but O-o-sa-ka... Tokyo and Kyoto are both on-yomi readings, while Osaka is kun-yomi.

Francis Drohan's book A Handbook of Japanese Usage has a nice little list of common おお words. - Sekicho 23:11, 17 March 2006 (UTC)

Apostrophe, I think

Which is the Nihon-shiki representation for タ, チ, ツ, テ, ト, ティ, トゥ, ツァ, ツィ, ツェ, ツォ? (Hepburn's ta, chi, tsu, te, to, ti, tu, tsa, tsi, tse, tso)

—Nethac DIU, would never stop to talk here—
23:48, 14 August 2006 (UTC)

Ta ti tu te to for the first five. The rest are not standard Japanese and are undefined even in Kunrei. Jpatokal 10:22, 15 August 2006 (UTC)

Move?

The title doesn’t seem quite appropriate for an article in the English Wikipedia. Wouldn’t ISO 3602 Strict, for instance, work better? The same applies to Kunrei-shiki Rōmaji, which would be ISO 3602 then. Christoph Päper 15:35, 26 July 2007 (UTC)

Kunrei and Nihonshiki are much better known under those names than by the ISO code — which is unsurprising, as the Japanese names predate the ISO label by nearly a hundred years. Jpatokal 02:41, 27 July 2007 (UTC)

Maybe, but it’s not like there was one intuitive canonical name or written form: Ni[pp|h]on[-| |]s[h]iki[ [R|r][o|ō|ô]ma[j|z]i| romani[z|s]ation]. Combining the most common spellings (on WP), revised Hepburn and US, proper titles might as well be Nippon romanization and Kunrei romanization. Christoph Päper 13:17, 5 August 2007 (UTC)

"Nippon romanization" gets 16 (sixteen) Google hits, vs. 1620 for "Nippon-shiki" and 635 for "Nihon-shiki". Jpatokal 16:48, 5 August 2007 (UTC)

I get different numbers, but the same order of magnitude, but to be fair you should have compared “Nippon romanization” to “Nihon-shiki Rōmaji”, i.e. the actual current title of this article (and exclude Wikipedia from the search, but add transcription variants if you want).[2] That number is just as small – whatever this tells anyone. Christoph Päper 01:47, 22 August 2007 (UTC)

Lossless mapping?

The article states:

it is the only system of romanization that allows lossless ("round trip") mapping

But don't the long 'o' sound in ookii (おおきい big) and the long 'o' sound in ou (おう king) both transliterate to 'ô'?

How can you know whether a romanized word containing 'ô' originally contained a 'o' (お) or a 'u' (う) as the second kana of the long syllable?

In this sense, this romanization method is not lossless.

Another method simply using 'oo' and 'ou' (e.g. toukyou no toori) is lossless, and arguably more common in Japan than the Nihon-shiki described in this article (one reason being that it is difficult to type circumflexes, the other is that this is the way words are typed in IMEs).

laug (talk) 05:41, 13 September 2009 (UTC)

If you're familiar with Japanese, it's possible to distinguish oo and ou by context. Oo is used in yamatokotoba words, ou for Chinese. Jpatokal (talk) 02:40, 14 September 2009 (UTC)

By context is the definition of 'non-lossless'.

Macgroover (talk) 10:04, 20 October 2009 (UTC)

I agree with laug and Macgroover. For example, the word "crown, throne" 王位 is written in hiragana おうい and the word "many" 多い is おおい　in hiragana; both are pronounced as long "o" followed by "i". Of course they can be usually distinguished by context, but without context, representing both words as "ôi" means loss of information.

Actually, as I have understood (correct me if I'm wrong), there are four different situations relevant to the transliteration ou and long o:

1. In words of Japanese origin, long "o" is usually written in hiragana by appending お to the syllable ending with the "o" sound: so long "o" is おお, "to" with long "o" is とお　etc.

2. In words of Chinese origin, long "o" is written in hiragana by appending う: e. g. おう, とう etc. The same method is e. g. also used for writing long "o" in the word どう ("dô", meaning "how"), which I believe is not of Chinese origin.

3. In words of Western origin, long vowels are usually written by appending a horizontal dash, so long "o" and long "to" would look in hiragana like おー, とー. However, Western words are usually written in katakana (katakana is, unlike Nihonsiki, in perfect 1-1 correspondence with hiragana, and is used in similar manner as italic font in Latin scripts), so long "o" and long "to" are actually written as オー, トー.

4. The hiragana う may also occur after a syllable ending with the "o" sound in the jisho-form of a verb. In this case, as far as I know, う is pronounciated as "u" instead of lengthening the preceding "o" vowel: e. g. さそう (pronunciation: "sasou", kanji: 誘う, meaning: to invite), おう ("ou", 追う, to chase)

So another lossy aspect of Nihonsiki is that dash in Western words is lost: as I understand, situations 1, 2, 3 should all be transliterated as "ô". Moreover, in addition to the fact that conversion to Nihonsiki is not lossless, automatic roundtrip conversion is unfeasible also because the conversion the other way round is not lossless: hiragana does not distinguish situations 2 and 4 but as I understand, in situation 4 transliteration "ou" should be used instead of "ô". For example, the hiragana おう are used to write both the word 王 (king) and the word 追う　(to chase) but 追う should be transliterated as ou (it is pronounced as two different vowels: "o" and "u"), while 王 should be transliterated as ô (it is pronounced as long "o").

So the sentence in the article should be rephrased, e. g. that Nihonsiki is the least lossy of romanization systems: it is nearly lossless, with the exception of mapping long vowels.

Jaan Vajakas (talk) 16:38, 15 July 2010 (UTC)

For the sake of completeness, there is one more situation (mentioned earlier in this talk page) that I must add to my list:

5. The hiraganas お and う may also occur after a syllable containing "o" in compound words like みのお.

Jaan Vajakas (talk) 17:11, 15 July 2010 (UTC)

It is possible to distinguish between 王/おう as ô, 追う/おう as ou and おお as oo, but it is unclear whether the various standards of Nihonshiki/Kunrei actually mandate this. Whether a long vowel is written as おう or オー is (AFAIK) not significant, both represent the same sound. Jpatokal (talk) 17:49, 17 July 2010 (UTC)

That the standards are unclear makes matters even worse for roundtrip mapping. Leaving the other examples alone, we still have the pair 王位 and 多い which are おうい and おおい in hiragana, respectively, but both are pronounced the same and transliterated "ôi", AFAIK, which makes roundtrip transcoding impossible.

"Roundtrip" would mean that it is possible to make two algorithms, one for converting kana to Nihonshiki, one for converting Nihonshiki to kana so that the composition of the two algorithms is identity transform, possibly with the exception of some insignificant details. I am not so sure about whether the difference between オー and オウ (or おう and おー) is insignificant but I believe my teacher would certainly consider it a mistake if I used the hiragana おうい for 多い or おおい for 王位. So it is impossible to decode a transliterated text consisting of the single word "ôi" back into kana.

A Nihonshiki->kana conversion would not be very practical anyway, though, as Japanese is normally written as a mixture of kanji, hiragana and katakana, not only one type of kana (only katakana or only hiragana). However, it seems that in the early days of computing, the JIS X 0201 character set was used which had only katakana characters in addition to Latin characters, so such an algorithm might have had some use then, if it were possible.

All in all, my main point is that I still insist on rephrasing the sentence about the losslessness of kana-Nihoshiki mapping. Jaan Vajakas (talk) 17:30, 26 July 2010 (UTC)

(undent) This was recently deleted by an anon. I've reinserted it, with an explanation that losslessness is possible if and only if care is taken to distinguish ô/ou/oo. If not, and the standard does not mandate that it is, it will be not possible to convert it back reliably. Jpatokal (talk) 03:47, 11 June 2012 (UTC)

"ei" and long "e"

How is long "e" represented in Nihonshiki and Kunreishiki? AFAIK, い following a syllable ending with "e" is pronounced as long "e" in modern Japanese, but the Kunreishiki article says that the name of Kunreishiki in Kunreishiki is "Kunreisiki", not "Kunrêsiki". Was "ei" earlier pronounced as a diphthong (e. g. at the time when Kunreisiki was created)? Or is "ei" still pronounced as a diphtong on some occasions? Jaan Vajakas (talk) 16:53, 15 July 2010 (UTC)

Kunreishiki is a mapping of script, not pronounciation: the kana えい become ei regardless of how they're pronounced. Jpatokal (talk) 17:44, 17 July 2010 (UTC)

The Nihonshiki article says about long vowels only that "Long vowels are indicated by a circumflex, for example long o is written ô, unlike Hepburn, which uses a macron." (I guess there is no difference between Nihonshiki and Kunreishiki in that respect?). The article should elaborate more on long vowels: why is "ei" preferred to "ê" in "Kunreisiki", while "ô" is preferred to "ou" in "Tôkyô" (correct me if I'm wrong)? If the standards are unclear, then the articles should state that and bring prominent examples of how they are used de facto. Jaan Vajakas (talk) 18:00, 26 July 2010 (UTC)

That sentences should be "The long vowels o and u are...", since ei is not treated as a long vowel by any of the romanization systems (Hepburn, Kunrei or Nihonshiki). It would be nice to elaborate on why this is so, but quite frankly I have no idea, it's just the way it is. Jpatokal (talk) 11:54, 27 July 2010 (UTC)

Because ええ and えい are not the same sound, I'd guess. — Preceding unsigned comment added by 68.191.42.175 (talk) 02:13, 11 June 2012 (UTC)

I do not agree with 68.191.42.175: I am not a native speaker, but I have understood that in most words (though not all, as I previously thought), えい is pronounced as ええ, just like おう is pronounced as おお in words of Chinese origin; this is confirmed by [3]. It would indeed be interesting to know the historical reasons why えい and おう are treated differently in transliteration. Jaan Vajakas (talk) 09:44, 1 November 2012 (UTC)

Supplanted?!

Yes, foreigners tend to use Hepburn romanization, which is near-phonetic in English, but I've yet to see a native Japanese using Hepburn. Based on totally subjective observation, to me it seems that Nihon-shiki may even outrank Kunrei-shiki in frequency of use among native speakers. — Preceding unsigned comment added by 213.163.40.100 (talk) 13:43, 15 December 2012 (UTC)

Well, we have to ask ourselves this question. What is the purpose of having Romanized Japanese? It seems to me of no use to native Japanese, who already have hiragana, katakana and kanji to write to each other. I've yet to see a native Japanese using Romaji to write period. The only time I've seen that happen was to try and teach foreigners the pronunciation of a Japanese word, or Japanese grammar. (Based on my own subjective observation of course.) From what I can tell, it's about 50/50 among native Japanese speakers; they use either kunrei/Nihon-shiki or Hepburn, and when they use anything other than Hepburn, they raise eyebrows with foreigners. If you asked me, I think much ado is placed on coming up with a "Japanese" style romanization with a "lossless, 1 to 1 relationship with kana," which only Japanese speakers (because they will have been taught) or hardcore Japanese scholars will use (when they become aware of it), when, in the end, most foreigners (for whom the system exists anyway right?) will use Hepburn or some variant. Those things considered, I think it a waste of time that Japanese students have to learn this system very few people use. I posit that learning kunrei/Nihon may actually hinder Japanese students in learning other languages that use the Roman alphabet, because they learn to pronounce Roman letters in a Japanese way (basically a different version of kana), and then they will have to learn yet new usages, especially when they learn English. I've taught English to Japanese students of all levels, and in my experience, it's hard to get them to pronounce English words in a way other than Japanese romaji. Anyway, yes, I would argue that on the whole, Nihon/kunrei-shiki style romaji has been supplanted by Hepburn Romanization. Try as the Japanese government might, with its ISO registration and its officialization, the defacto Romanization of choice for the Japanese language is, in fact, Hepburn style. Nobody actually uses it, except maybe Japanese natives (to whom it is redundant) and/or snobby scholars of Japanese.Solar3939 (talk) 07:29, 10 March 2016 (UTC)

Use with Extended Katakana

Is all extended katakana undefined in Nihon-shiki? If so, some mention of this somewhere in the article would be appreciated. Rod Lockwood (talk) 08:57, 15 February 2015 (UTC)