Behind you! Behind you!

The modern word for "tree" is ki, but a lot of the compounds involving ki use ko instead: kodama ("echo", literally "tree-spirit") and kogarashi ("cold wind", literally "tree-afflicter") are two commonly-occuring examples.

Similarly, the modern word for "fire" is hi, but we find compound words like honoke ("[the feeling of] fire's presence", or maybe "smoke"; literally "spirit of fire") and hokage ("firelight").

(Both of these is are actually ïs in OJ, which may be relevant -- hi as in "sun" or "day" had a regular i, and I don't believe it has been shown to act like a ho in any ancient documents.)

This being the case, it really should not have been as surprising to me when I learned that ushiro ("behind") was related to shiri ("rear end" or "ass"). The u- is apparently related to modern mi (身, "body", "contents", etc.), in that both are derived from an older form mu -- which survives in words like munashii ("empty", "lonely", from mu + nashi ("absent")). Apparently, in a compound word like this, mu turned into a u, presumably via the same process that gave us words like ikou and Kobe, but as a standalone word it turned into mi.

If all of this is correct, then ushiro is a neat linguistic fossil, preserving evidence of not one, not two, but three of Japanese's major sound shifts.

Popularity factor: 8


Funny, I've always assumed that the hi of 日 and 火 were originally the same word. Are you saying this isn't true?

Also, doesn't it strike you that the "o" on the end of ushiro and in, say, hokage are different? In the latter, it's the sound change that occurs in a compound, whether because the linkage protects a sound usually worn away when in isolation, or because of some now invisible linking particle that causes the ablaut. I found a fascinating Japanese site once that claimed the latter, and held it to be the 定説 no less. I'll search for it after work today.

Another interesting thing on the same site was a phenomenon of "extended roots", whereby related new words were produced from smaller stems with the extension of a -k- or an -m- or a -t-, ending the syllable with a necessary "i". Thus pairs like "kata/katachi", "tsu/tsuki" (as in "tsugomori" "new moon"), and so on. Perhaps the "i" in "shiri" is another case of this, and there's a more basic "shi" root here? Maybe related to the "su" in "sugata"?


I mistyped.

First sentence: ...the "hi" of...

Second paragraph, ...on the end of "ushiro" and in, say, "hokage"...



I always assumed they were the same word, too! and I still wouldn't be surprised if I learned they shared the same root, but apparently in the manyougana they are regularly distinguished: "fire" is hi2 and "day" is hi1. That means that they were probably pronounced differently and for all intents and purposes were different words. (It would amuse me in a nerdy way if they started as the same word, diverged into two separate words in the 8-vowel system, then reconverged again in MJ.)

I'm interested in that site you mention, because most of my sources claim that "ho", "ko", etc. were older forms preserved in compounds rather than variants created by invisible linking particles. I can see how the latter theory makes sense, especially since in my (admittedly limited) reading I don't recall ever coming across "ho" or "ko" used as standalone words (except in iffy maybe-a-compound, maybe-a-phrase "no" structures like "konoha" (木の葉) or "honoho" (炎)). On the other hand, if the "i2 -> o, i1 -> i" pattern is regular (I've only investigated it for "ki" and "hi", and "shiri" is unknowable because the dual vowels are not observed after "r"), divergent older forms would make more sense to me. So please, if you can find the link, post it!

As to whether the "o"s in "ushiro" and "hokage" are different-- yeah, you're right, the big issue is that "hokage" has the "o" before a compound and "ushiro" does not. If the invisible particle theory is correct, this makes it unlikely that they were produced by the same process.

There was another word meaning "behind", though, which was "shirihe" (後方) (in opposition to "mahe", i.e. modern "mae", presumably derived from the "ma" in "massugu" etc.). Maybe "(mu)shiro" was originally "(mu)shirohe", and dropped the final sound for some reason? (This is a pretty far-fetched theory in the absence of any textual evidence, I know, since it requires a final mora to disappear AND the word "shirihe" to be recreated after the "o" form had disappeared. Not impossible, but very rickety as a hypothesis.)


Um, I've got a problem with mu changing into au into ô in ikou. (I'll have to dig around a bit for Kobe...what's the hypothesized original form?)

In the verbs, it was stem/ending -a + mu, with later elision of the m (and presumably compensatory nasalization) bringing the a and u together, right?

Not to be stroppy, just 気になった。


D'oh! That's a formatting error on my part, there should be a space in there. "/mu/ turning into a /u/" (via the process you describe). Sorry!

According to the link the Koube derives from Kamibe, but actually I think it came directly from Kamube. Kamube, Kanbe, Kaube, Koube. (Because kamu:kami::ko:ki, whether it's because of older forms or ablautification. It's just less obvious because of the mu->n change that gave us words like "kannushi" and "kannagi")

Actually, speaking of that, you know what word would provide a good second data point backing up "mu + shiro -> ushiro"? "ume". That was originally "mume", right? mume, nme, ume; mushiro, mshiro, ushiro.



The promised page, which itself links to a number of other juicy pages on Japanese historical linguistics (omake ni).

Seems I had it remembered wrong, the real 定説 is the former, that the compound forms are actually the older ones, here protected from a nominalizing particle い that seems to have spread like a bad cold among nouns of all kinds of vocalic persuasions. い itself appears in the 万葉集, and lasts even today in あるいは, according to the article. The author goes a step further and links this い particle to the similarly nominalizing い of the 連用形, which makes sense, though a little too much sense to accept easily, if you know what I mean.

The thing I find intriguing is that the a>e pairs are so very more common than the others. I guess one could just say that roots in "a" were more common to begin with, but that's not very satisfying, either. Off the top of my head, the only active words in my vocabulary that go o>i are those you mentioned yourself, like hotaru(火垂る) and kodachi(木立ち). Further search online finds an English article (ling.cornell.edu/Whitman/ VowelsofProto-Japanese.Martfest.pdf) that further lists よみ>よもつかみ, where よみ means something like the underworld, related to 闇.

And by the way, on an unrelated note, did you know that Ono Susumu, the editor of the いわなみ古語辞典, actually thinks that Japanese comes from Tamil?! A week of studying Korean should convince anyone that the origins of Japanese obviously lie in another direction, but...


Thanks! I'll check all those out once I'm back from Osaka.

I did know that about Ono.. I really don't know what to think about it, though. He is an incredibly well-read man when it comes to old Japanese, and he seems to be the source of a lot of newly discovered connections between Japanese and Korean, so he's not just some Tamil-obsessed crank... on the other hand, if his evidence is that persuasive, why isn't the theory more widespread? I'll get around to reading the book one day, I guess.

(In fact, one of the J-K things in the last edition of his dictionary was the assertion that 蛍 is not from 火垂る, but rather related to a Korean word "pontari" that also means "firefly". This came as a great shock to me.)


PROFESSOR WHITMAN! Hell ya! Represent!

Comment season is closed.