Shanghai is Shanhai in Japanese / しゃんはい

However, 深海 is shinkai (shēnhǎi in Chinese).

Why does it seem like the Japanese switched the letter "h" to "k"? I know the kana, and "h" would go to a "b" sound, "k" would go to a "g" sound. I am just not understanding why they would say "shinkai" (if not "shenkai") and not "shinhai" in the above example.

  • Comments have been moved to chat; please do not continue the discussion here. Before posting a comment below this one, please review the purposes of comments. Comments that do not request clarification or suggest improvements usually belong as an answer, on Japanese Language Meta, or in Japanese Language Chat. Comments continuing discussion may be removed.
    – chocolate
    Feb 5 at 3:39
  • @chocolate Was that for me or another user? I am unclear why comments had to be moved. It doesn't show what comments were moved. Feb 5 at 19:25
  • When the comment section gets long we often move it. It has nothing to do with you personally.
    – A.Ellett
    Feb 5 at 20:38
  • I voted to close. I think you need to clarify what you are after. Is it how the Japanese reading for 海=kai developed historically or why the modern Japanese reads 上海 as shanhai and 深海 as shinkai? In case of the former, the existing answer should answer the question; in case of the latter, it just because 上海 is a place name and the modern Chinese pronunciation sounds that way to Japanese ears (mostly). There's not much 'why' here. If you think I misinterpret your question, you would better provide more examples to clarify the question.
    – sundowner
    Feb 6 at 12:25
  • 1
    @sundowner - This poster is confused beyond help as you can see from how the chat went. (I was tempted to say the word 電話 was also ... but refrained from it...) The post, meanwhile, has gotten its maximum value with its edited title and informative answer. Let's keep it closed.
    – aguijonazo
    Feb 7 at 1:34

1 Answer 1


h in romanization of Mandarin language is [x] (and e.g. in Cantonese it is [h]).

[h] is recently developed consonant in Japanese language itself:

Old Japanese (8th century) and some earlier part of Middle Japanese (which was period generally about 800 - 1600) had [p] and [ᵐb] (i.e. prenasalized, see here). (Prenasalization of b, d, z, g existed until at least 17th century.)

Intervocalic [p] changed to [w] in second half of 10th century (according to Frellesvig (2010)), and later, many remaining [p] changed to [ɸ] (which is different from [f] but in romanization of Japanese usually romanized as f).

[w] (both from original [w] and from [p]) before /i/, /e/, /o/ was lost in various stages in 10th - 13th centuries.

は, ひ, ふ, へ, ほ were still /ɸa/, /ɸi/, /ɸu/, /ɸe/, /ɸo/ at least in 17th century (or in native words in non-initial positions they represented /wa/, /i/, /u/, /je/, /o/).

Probably somewhen in 17th - 19th century /ɸa/, /ɸe/, /ɸo/ started changing to /ha/, /he/, /ho/, and /ɸi/ started changing to /çi/. Pronunciation with [ɸ] still remains in some dialects, e.g. Izumo dialect (one of Unpaku dialects).

Bjarke Frellesvig (2010, "A History of the Japanese Language", pages 386-387):

14.3 Delabialization of /f/; /f/ > /h/

The final change to affect the reflex of OJ /p/ was the change of /f/ (see 11.3) to /h/ which occurred at some point during NJ. This change was primarily a phonetic change with no mergers or other immediate phonemic impact and it was therefore not reflected in general writing. Thunberg (1792) mostly writes words reflecting LMJ /f/ with 'f' (e.g. fanna 'flower'), but says that words with 'f' sometimes are pronounced ('read') with 'h' (e.g. hanna) and Siebold (1826) uses 'h' in his transliteration of the kana letters はひふへほ, but notes that they are often pronounced with 'f'. However, by the middle of the nineteenth century, sources in alphabet writing generally write 'h', except before 'u' where it is 'f', reflecting that /h/ is pronounced [ɸ] before /u/. Hepburn's transcription system from the end of the nineteenth century adopted this allophonic notation which since then has dominated transcriptions in the west. Very recently, due to a large intake of loanwords from English, [ɸ] has become phonemic /f/ before vowels other than /u/ (14.6).

(Bjarke Frellesvig's book uses different phonemic analysis than my analysis. I prefer to use /ɸ/ instead of /f/ in phonemic notation of Japanese.)

[x] and [h] sound similarly, but when Japanese language had neither of them, the consonant sounding most close to [x] or [h] (e.g. in words in Chinese languages) amongst available consonants in Japanese language was [k], and therefore [k] was used during borrowing (e.g. from Chinese languages). [x] and [k] have the same place of articulation, but different manner of articulation.

When words are borrowed from one language to another language, and source language contains sounds not existing in target language, only very rarely those sounds are introduced in target language. Known historical example is that ancient Romans in first century BCE introduced letters Y and Z at the end of Latin alphabet to write words borrowed from Ancient Greek, which had /y/ and /d͡z/ ~ /zd/, later /z/ (discussion on pronunciation of zeta here) in Ancient Greek. But less educated Romans were probably pronouncing Y as [i].

  • 1
    Great answer. Just hope it is appreciated by OP. So the short answer is our ancestors couldn't distinguish /k/, /kʰ/ and /x/? This Wikipedia article lists 見, 溪 and 曉, respectively, as examples for those three initials. I wondered how 曉 ended up with /g/ but it turned out it's きょう in both 呉音 and 漢音 and ぎょう is 慣用音.
    – aguijonazo
    Feb 3 at 21:38
  • 1
    @aguijonazo Maybe some scholars, monks, nobles etc. received proper education in Middle Chinese and could speak it properly, but it is possible that even they, when using Chinese-origin words in Japanese sentences, were using sounds assimilated to Japanese sound system. And persons without education in Middle Chinese may even have been unawere of /kʰ/, /x/ etc. if they only saw written text. If Japanese government wanted to more faithfully preserve Chinese sounds in borrowed words, then government would have to extend hiragana, katakana...
    – Arfrever
    Feb 4 at 2:16
  • 2
    @watashinopiza You're confusing things by bringing up Shanghai once again. We've explained already why place names may not follow the more traditional readings.
    – A.Ellett
    Feb 5 at 20:41
  • 1
    @watashinopiza Japanese would have pronounced 上海 with a k sound now if they took the name from the Sui/Tang dynasties. But they didn't, they took the name from Mandarin. Let's use another example: 長安 is rendered as Chou'an in Japanese, because they imported the name from antiquity. If they imported the name from Mandarin, it would be Chan'an.
    – dROOOze
    Feb 5 at 22:22
  • 2
    @watashinopiza Your question starts with the matter of the pronunciation of Shanghai in Japanese vs Chinese. You've brought the subject up of Shanghai over and over again. You keep making this about the name of a place "Shanghai". You yourself have made this all about place names. If it's not about place names, then stop bringing up the matter of Shanghai--it is quite literally a exception to an otherwise fairly consistent rule in Japanese.
    – A.Ellett
    Feb 7 at 13:56

Not the answer you're looking for? Browse other questions tagged .