Chris Kern has an entry on “ghost characters”:

Proving once again that the Japanese writing system is supremely screwed up, there are apparently certain characters called 幽霊文字 (“ghost characters”) that have no readings, meanings, or examples of use. Even if you look them up in a dictionary you get definitions like 意義未詳 (reading and meaning unknown). Examples of these ghost characters are 暃 and 碵.
They all come from the JIS set, which is a set of characters that are standard for computer terminals to display. Apparently during the compliation of the JIS set, some characters that weren’t actually characters got onto the list accidentally—either because they were miswritten versions of actual characters or the compilers misread certain kanji.

Matt of No-sword, in his post on the subject, shows that some of the characters are real, if obscure, but adds “even the JIS bigwigs admit that 妛 and 椦 are indeed just mistakes.” Something of a parallel to ghost words in English.


  1. Some superrant and hownely shindled observations there, LH.

  2. No-Sword’s entry was interesting; some of those ghost characters aren’t quite dead yet, being used in unique place names.
    I believe there are also a small number of Japanese names that have *silent* kanji. If you’ve studied Japanese, you know just how weird that would be.

  3. Silent kanji? Now I’ve heard everything. Matt, do you have anything to say about this?

  4. John Emerson says

    There are a number of rare Chinese characters used only in proper names whose actual pronunciation is uncertain. One is p’i/p’ei, the given name of Emperor Ts’ao P’i of the San Kuo Wei dynasty. Another is sou/shu, the given name of Liang Sou-ming, subject of the book “The Last Confucian”. Unfortunately I can’t find the Chinese graphs from here.
    Some Chinese dynasties deliberately gave their children obscure names to avoid the nuisance of requiring th taboo of common characters. For example, early Han emperors were named hsuan“dark”, cheng “right”, and pang “state”, requiring that Han texts of the Tao Te Ching be garbled because of taboos.

  5. Yeah, it’s true, there are some words that include silent characters. Mostly names. One that you still see a lot is 右衛門 (some kind of palace guard) — it can be and was originally pronounced “u.e.mon”, but just “(0).e.mon” is much more common. Obviously this happened because the initial “u” (meaning “right (not left)” was dropped, but the result is that 右 is left without a sound to call its own.
    右衛門 was a common element in names, so it’s still floating around in restaurant names and elsewhere, silent 右 and all. (For example, there’s a popular brand of bottled green tea called 伊右衛門, which is pronounced “i.(0).e.mon”)
    Here’s another good example of high weirdness in the Japanese writing system. One of the exits at Ueno station is called “Shinobazu-guchi”. “Guchi” just means “exit” (literally “mouth”) so the name is “Shinobazu”. The actual etymology of the name is AFAIK unclear, but, “shinobazu” happens to be exactly the way you would say “does not hide” or “does not endure” (if you were using the old “-zu” negative ending instead of “-nai”).
    But it’s written like this: 不忍, Chinese-style, where the negative comes BEFORE the verb (instead of after as in Japanese). So the first character, 不, carries the meaning of the “azu”, and the 忍 carries the “shinob”, and it’s impossible to rationalize the characters with the sounds one-by-one in order. If someone said to you “how do you pronounce the first character in the place name 不忍”, the only possible answers are “you can’t pronounce only that character” or “well, ‘azu’, but you pronounce it AFTER you pronounce the second character”, neither of which gel with most people’s understanding of how writing works. (This kind of tomfoolery with 不 and “-azu” was actually scarily common back when Chinese writing was still influential in Japan — it would appear in the middle of otherwise normal Japanese sentences.)
    I mentioned this here:
    and there was a bit of interesting discussion. Like Amida says, ‘Chinese characters are not “ideograms,” as they are often called, but no one seemed to have told the ancient Japanese that!’

  6. But we know English knows enough not to have silent letters.

  7. Now that my brain has been de-Sinocentrified, I see how the “shinobazu” example above goes through the intermediary step of Kambun, in which diacritical marks tell you how to read Chinese as Japanese. 不忍 is Kambun without the diacritical marks. Japanese just gets “curiouser and curiouser” for me….

  8. Perhaps urban legend, but I’ve heard that map makers often put planned minor errors into their maps so that they can make sure whether or not other map makers are stealing their material. Perhaps there’s something to do with those English “ghost words” and Webster? Why not just remove the ghost words if they’re not actually content that the dictionary cares to have in it?

Speak Your Mind