6

For example:

  • て＋つな＝たづな（手綱）
  • め＋ふた＝まぶた（瞼・目蓋）
  • かね＋つち＝かなづち（金槌）

The only thing I can see for sure is that the second word becomes voiced, but that's more of an after-the-fact thing than a rule that dictates when the sound actually changes from e to a.

Is there a rule or pattern to it?

ssb
  • 18,213
  • 2
  • 54
  • 95
  • 2
    Possible duplicate(?):[Why is a place that sells さけ a さかや?](http://japanese.stackexchange.com/questions/3533/why-is-a-place-that-sells-%E3%81%95%E3%81%91-a-%E3%81%95%E3%81%8B%E3%82%84), [Exceptional compounding forms](http://japanese.stackexchange.com/q/2534/542). – Flaw Nov 06 '12 at 08:47
  • These are also helpful, thank you. Didn't know how to go about searching this one. – ssb Nov 06 '12 at 08:53

1 Answers1

3

e does not become a. Rather, it is the other way around: a becomes e. More specifically, there are two forms of e: e1 and e2. (See 上代特殊仮名遣) The rule is a + i > e2. Both e1 and e2 merge into e after Nara period.

For reference, these pairs are termed 被覆形 and 露出形. The form without the -i suffix is 被覆形 while the -i suffixed form is 露出形.

Dono
  • 11,177
  • 7
  • 51
  • 63
  • 1
    [Here's a list of examples.](http://jbbs.livedoor.jp/bbs/read.cgi/study/10958/1283085223/18) – jogloran Nov 06 '12 at 05:49
  • 1
    You mention that *a+i > e2* > Modern *e*, and that the original root ends in *a*. What is the *i* which yields modern *e* when the word occurs free and not in a compound? – jogloran Nov 06 '12 at 05:50
  • That explains the surname 金井 and others, too. For some reason I didn't even connect this in my mind to other similar transformations with other sounds. Thank you! – ssb Nov 06 '12 at 06:13
  • @jogloran It is sometimes said to be some kind of subject or topic marker. – Zhen Lin Nov 06 '12 at 23:43
  • @ZhenLin The subject / topic marker -i is usually considered to be separate. Also, due to the skewed distribution, it is argued to be loan from Old Korean. There are some fossilized forms surviving such as arui (wa). The -i mentioned here with nominals is more likely to be *C > *y > i, where C is some consonant that lenited. – Dono Nov 07 '12 at 00:11
  • The compound words all have the structure *X of Y*, but it seems unlikely that any historical genitive particle like *tsu* or *ga* could have yielded a vocalic *i*. Incidentally, I wonder if there's any evidence linking *tsu* to the fossilised Korean genitive particle *s* (e.g. 바닷가 *bada-s-ga*)... – jogloran Nov 07 '12 at 00:38
  • @jogloran Regarding -tu and Middle Korean -s (Old Korean -ci), Martin proposed the etymology in "Morphological clues to the relationship of Japanese and Korean". For a comprehensive up-to-date analysis and review of the evidence, I suggest that you read "A Descriptive and Comparative Grammar of Western Old Japanese" (part 1) pages 152-158 and "Koreo-Japonica: A Re-evaluation of a common genetic origin" pages 49-53, both by Alexander Vovin. – Dono Nov 07 '12 at 11:40