Consider the following words:

  • 燦々 sansan
  • 散々 sanzan

Both of them are formed via the reduplication of san, but the latter word has the second instance voiced, while the former doesn't.

Was there some sort of historical phonological rule that governed whether or not the second syllable would be voiced, or is this property of reduplicated words unpredictable a priori?


Sequential voicing (called 連濁【れんだく】 in Japanese) isn't predictable, but there are rules that describe when it's "blocked"--in other words, when it's much less likely to occur. None of these rules are absolutes, though, and we can find some exceptions.

Reduplication commonly results in sequential voicing:

  • 黒々    (くろぐろ
  • 人々    (ひとびと
  • 華々しい  (はなばなしい)
  • それぞれ  (それぞれ
  • 口々    (くちぐち

This process is blocked in reduplicated onomatopoeia, which has its own rules. Basically, both halves should sound the same, regardless of whether a sound is voiced or not:

  • とんとん
  • どんどん
  • ぴくぴく
  • ひくひく
  • びくびく

And it's blocked when the reduplicated portion already contains a voiced obstruent. (This is just a restatement of Lyman's Law as applied to reduplication.) So:

  • 度々    (たびたび)
  • 飛び飛び  (とびとび)
  • 数々    (かずかず)
  • 繁々    (しげしげ)

This process is also usually blocked with Sino-Japanese compounds, which is to say, those compounds that use on readings:

  • 空々    (くうくう)
  • 個々    (ここ)
  • 半々    (はんはん)
  • 少々    (しょうしょう)
  • 戦々恐々  (せんせんきょうきょう)
  • 燦々    (さんさん)

There are exceptions, though, where reduplicated Sino-Japanese compounds are voiced, so it's not an absolute rule. These are all the exceptions I'm aware of, although there are undoubtedly more:

  • 甲斐甲斐しい (かいがいしい) [but this may not be S-J at all, see comments]
  • 散々     (さんざん
  • 種々     (しゅじゅ
  • 精々     (せいぜい
  • 騒々しい   (そうぞうしい)
  • 猩々     (しょうじょう

In certain cases this can be explained as a property of the word being reduplicated. For example, 甲斐【かい】 voices very readily in compounds:

  • 不甲斐ない  (ふがいない)
  • 生き甲斐   (いきがい
  • 甲斐甲斐しい (かいがいしい)

So it's no surprise that it voices in our reduplicated example. Why does this word voice when others do not? It may be that this word is treated as a nativized part of the Japanese vocabulary, rather than as a Sino-Japanese compound, making it subject to the phonetic rules that normally apply to native words. And the same explanation can be used for other Sino-Japanese compounds that readily undergo voicing, such as 会社:

  • 株式会社   (かぶしきがいしゃ
  • 信販会社   (しんぱんがいしゃ
  • 親会社    (おやがいしゃ

So we can see that in some cases, our exceptions are because of a lexical property of the word that undergoes voicing.

In other cases, though, I have no explanation. Why is 猩々 voiced while 少々 is not? Why is 散々 voiced while 燦々 is not? It's possible that the same explanation applies, but I find it much harder to support. After all, 種 never voices to じゅ in compounds other than 種々.

So I think the best explanation is that the rules "blocking" sequential voicing are merely strong tendencies rather than actual rules. What's more, in cases where it's not blocked, whether or not it occurs isn't predictable. And as a result, we can make some generalizations about when sequential voicing occurs, but we can't predict it with certainty. I'm afraid you'll just have to memorize cases like 燦々 and 散々.

  • 5
    There seems to be reason to believe that 甲斐 is native Japanese. FWIW: oshiete.goo.ne.jp/qa/576294.html
    – dainichi
    Oct 9 '13 at 21:42
  • @dainichi Oh, thank you for the link! I had no idea. If that's so, then of course it's not an exception at all.
    – user1478
    Oct 9 '13 at 21:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.