Japanese Romanization: they still haven't decided, part 2

« previous post | next post »

For a country that already has Chinese characters (kanji) and two syllabaries of its own (hiragana and katakana; see also furigana), judging from the ubiquity of romaji across the country, it would appear that they are well into the process of turning Latin letters into an integral component of their quadripartite writing system.  Some may argue that they already have done so.

What's going on?

Why hasn't something similar yet happened in China (Vietnam's writing system is already clearly based on the Latin alphabet)?

"Akasi or Akashi? Hepburn Most Established of Japan’s Different 'Rōmaji' Systems", Nippon.com newsletter (11/2/22)

Signs on highways and at railway stations in Japan show place names in both Japanese and Roman letters, although the rōmaji system employed can vary. The Hepburn system dominates, but the Kunrei and Nihon systems are also seen around the country.

Some excepts:

Japanese words spelled in rōmaji, or the Roman alphabet, are ubiquitous in Japan, appearing on signage on roads and at train stations as well as on product labels and in advertisements. Elementary school children first learn to spell Japanese words using the Kunrei system, but the Hepburn system is the most widely used. A less well-known Nihon system also exists.

A survey on Japanese language use by Japan’s Agency for Cultural Affairs sheds light on the prominence of the different systems. Concerning the Romanized spelling of the Hyōgo Prefecture city 明石, 75.4% of respondents chose the Hepburn spelling “Akashi” and 23.3% the Kunrei spelling “Akasi.” Hepburn was also the dominant form when rendering the prefecture 愛知, with 88.0% spelling it “Aichi” compared to 10.8% who chose the Kunrei form “Aiti.”

The various rōmaji styles have distinct backgrounds. The Hepburn system was devised by a group that included US missionary James Curtis Hepburn, to whom history has assigned sole naming credit after he adopted the form for the third edition of a popular Japanese-English dictionary, published in 1886. The Kunrei system derives from the earlier Nihon system and is the first form that most children learn. As Hepburn is easier for people with English-language backgrounds to pronounce, it has become the dominant form and is used on important documents like passports and road signs.

The survey also looked at the frequency people used rōmaji when typing Japanese. Respondents in their twenties, thirties, and forties, who rely heavily on digital devices like computers and smartphones in their work and personal lives, used it most often. Overall, 54.7% of people said they used it either regularly or occasionally, although many said that the existence of three overlapping systems resulted in some confusion regarding spelling.

Jim Breen remarks

I rather liked the "sole naming credit" reference to Hepburn.

The article — both in texts and in charts — has some interesting information on frequencies of use of the different Japanese romanization systems.

From the Rōmaji nikki (Romaji Diary) to today's Rōmaji chalkboards in cafes, and from digital inputting to transportation signs and advertisements, we see romanization playing a variety of roles in Japanese literature and daily life. It's hard to predict where it will end, but it's fun watching the transformations.


Selected readings



  1. Antonio L. Banderas said,

    November 8, 2022 @ 8:01 pm

    In July 2020, 2.6 billion people (36% of the world population) use the Latin alphabet.

    It is extremely rare in Japan to use it [rōmaji] to write Japanese (except as an input tool on a computer or for special purposes like in some logo design), and most Japanese are more comfortable in reading kanji and kana.

  2. Jim Breen said,

    November 8, 2022 @ 9:12 pm

    Many years ago we stayed in a friend's house in the countryside near Matsumoto in Nagano prefecture. We used the little private 新島々 train line to travel to/from Matsumoto. That line used Kunrei romanization on its signs, so I had to explain to my family why we were going towards "Sinsimasima".

  3. Philip Taylor said,

    November 9, 2022 @ 5:02 am

    I don't have the necessary skills to identify which system of Romanisation Google Translate uses for Japanese, Jim, but I did find it interesting that it renders " 新島々" (unspaced) as "Shinshimashima" but " 新 島々" (spaced) as "Shin shimajima".

  4. J.W. Brewer said,

    November 9, 2022 @ 10:50 am

    @Philip Taylor: That's a glitch of some sort that does not reflect a choice (even an internally inconsistent choice) between romanization systems. As I recently read someone say in another context, because AI-type systems seem to emulate what humans do but do so in extremely non-human ways, they can occasionally fail in bizarre ways that a human would not.

  5. DaveK said,

    November 9, 2022 @ 7:17 pm

    Side issue, but what is the logic behind calling it the Latin alphabet versus the Roman alphabet? It just seems to introduce a (slight) inaccuracy since Latin is only one of the languages written in the alphabet and there are some differences depending on which language it’s used for.

  6. John J Chew said,

    November 9, 2022 @ 8:54 pm

    When one of my sons was born and we wanted to register his birth here in Canada with a Japanese middle name, I discovered that ‘ū’ is not a permitted character in names here, so we had to replace it with “uu”.

  7. John J Chew said,

    November 9, 2022 @ 9:01 pm

    I sometimes wish that English had furigana for toponyms. In Toronto, where I live, street signs are often surtitled in Chinese or Greek according to whatever the predominant neighbourhood language is. This is how I learned how to pronounce the name Gough (or as the Greek helpfully put it, ΓΚΟΦ).

  8. Philip Taylor said,

    November 10, 2022 @ 7:40 am

    I thought it interesting, John, that Greek requires both gamma and kappa (in that order) to represent the sound of a hard English "g" — I must have already been aware of this without it ever impinging on my stream of consciousness. But it also reminded me of something that I learned only in the last 24 hours — that the Mandarin Chinese word for "brother" 哥哥, which is spelled "gēgē" in pinyin, if expressed in IPA is not (as one might have thought) [ˈɡɜ gɜ] but rather [kɯ̯ʌ kɯ̯ʌ] — a rather complex and unexpected sequence for what seemed, to my untrained ear at least, to be a rather simple sound.

  9. Philip Anderson said,

    November 13, 2022 @ 1:17 pm

    @Philip Taylor
    That’s in Modern Greek, where G, B, and D are foreign sounds (gamma, beta and delta having become fricatives). In the combination above, gamma represents /ng/ – cf ‘nt’ for /d/ and ‘mp’ for /b/ (as in μπύρα beer).

  10. Michael Watts said,

    November 13, 2022 @ 5:24 pm

    I thought it interesting, John, that Greek requires both gamma and kappa (in that order) to represent the sound of a hard English "g" — I must have already been aware of this without it ever impinging on my stream of consciousness.

    /g/ was a part of Ancient Greek and written with gamma. Sound change replaced /g/ with a much weaker sound, but /g/ is (still? once again?) available in modern Greek, and it has to be written somehow.

    I have read that when the same thing happened in Old English, and then /g/ was imported (from Norse?) in words like "give", the spelling wasn't changed, but I have to say I prefer the Greek way of spelling different sounds differently.

    My impression from listening to Chinese singing has not been that pinyin "e" represents a diphthong. But I guess I could be wrong about that.

  11. Philip Anderson said,

    November 15, 2022 @ 8:50 am

    @Michael Watts
    English spelling is notoriously conservative, generally not reflecting sound changes, and keeping the spelling of words borrowed from other languages.
    But spellings with initial ‘g’ are different from the Greek example. In OE, it remained /g/ before back vowels (go, gang, gut) and was only palatalised before front vowels – but these words were later re spelled with ‘y’ (yet). So using ‘g’ for ON borrowings did not create ambiguity. The problem came when words were later borrowed from French and Latin etc, which gave us ‘g’ pronounced as /j/ before a front vowel.
    As for place names, England has Gillingham in Kent with /j/, and in Dorset with /g/! But I don’t think we can blame the Vikings.

  12. Philip Taylor said,

    November 15, 2022 @ 1:08 pm

    "Gillingham in Kent with /j/" — as a Kentish Man, I dispute that ! Gillingham in Kent starts (phonetically) with a /dʒ/ (/ˈdʒɪ lɪŋ əm/).

  13. Michael Watts said,

    November 15, 2022 @ 9:05 pm

    In OE, it remained /g/ before back vowels (go, gang, gut) and was only palatalised before front vowels – but these words were later re spelled with ‘y’ (yet). So using ‘g’ for ON borrowings did not create ambiguity.

    But the ambiguity is present in Old English. Students of Old English are warned about it, and a diacritic mark is used on the letters c and g to tell the modern student whether the letter should be given a native or borrowed pronunciation. The respelling with y, as you note, happened at a later stage.

    I'm contrasting this unfavorably with modern Greek, where borrowed sounds are given spellings that unambiguously indicate their pronunciation. Old English took the opposite approach.

    Philip Taylor, I think Philip Anderson was just confused about what sound is represented by /j/. It would make no sense to say that borrowings from French gave us "g pronounced as /j/ before a front vowel".

  14. Philip Taylor said,

    November 16, 2022 @ 5:54 am

    Yes, of course, I clearly overlooked that at a first reading. "‘g’ pronounced as /j/ before a front vowel" should have triggered warning bells, (but failed to do so), while a putative /ˈjɪ lɪŋ əm/ just jeapt off the page (so to speak), probably because a concrete example such as "Gillingham" is faster for the lay reader to take in than an abstract "pronounced as /j/ before a front vowel"

  15. Philip Anderson said,

    November 17, 2022 @ 9:16 am

    @Philip Taylor
    I can’t write ʒ on my phone except by finding one to copy, which is sometimes too much effort!

  16. Philip Taylor said,

    November 17, 2022 @ 12:40 pm

    Ah well, I have never needed to type a ʒ in order to make a telephone call, so I can only sympathise …

    But returning (briefly) to "Gillingham", I am now reasonably certain that a better phonetic transcription would have two /l/s, one to terminate the first syllable and one to start the second, as in /ˈdʒɪl lɪŋ əm/.

  17. Gui said,

    November 19, 2022 @ 7:23 pm

    For me, the Kunrei system is kind of crazy. It's adapting Latin letters to imitate Japanese sounds but then ignoring the pronunciation rules that govern the letters. How could anyone be in favor of this? It would be like English speakers using kanji to imitate English words but then changing the pronunciation of the kanji. Wha?!

    Because of this, when Japanese experience Kunrei as their introduction to Latin alphabet, they learn incorrect pronunciations that dog them into later school years. Another pet topic of mine is Japanese children being taught to hand-write the alphabet following a font only used in typing. I constantly have have to re-teach the letter "a" which kids always diligently and slowly trace out exactly as it appears in this comment.

RSS feed for comments on this post