56

When you first begin to learn Japanese you are taught that Japanese has no stress and each syllable should be pronounced equally.

You also learn that certain vowels are not pronounced, or only pronounced very slightly, such as the "u" in "desu" and the "i" in "deshita".

But it seems that sometimes these vowels are pronounced if they occur in the first syllable of a word such as "sugoi", "subarashii", "shiro", "shimbun".

Is this a kind of exception to both rules or is it just something a foreigner might think they hear which is not really there?

hippietrail
  • 15,317
  • 30
  • 98
  • 146
  • 16
    "[...] Japanese has no stress and each syllable should be pronounced equally." I really hope this fallacy dies someday. The idea that Japanese has no stress sounds like a stopgap fix to prevent English-speaking students from mangling the pronunciation by forcing stress accents onto what should be high-low intonation. – Derek Schaab Jun 14 '11 at 14:09
  • 6
    @Derek: Japanese indeed has no stress accent in the English sense. But it does have an accent (high-low, as you've mentioned). I guess what you've really meant is that Japanese doesn't have a stress accent, but it does have stress? – Boaz Yaniv Jun 14 '11 at 18:37
  • 3
    @Boaz: Right, there seems to be a misconception in a lot of Japanese classes that Japanese has "no accent at all" and should be pronounced "flat". Which of course isn't what you hear if you listen to actual Japanese. The differences between stress accents and high-low accents gets glossed over all too often with these convenient untruths. – Derek Schaab Jun 14 '11 at 19:33
  • 1
    @Derek: I've actually only heard it in very few places (then again, the only Japanese classes I ever went too were taught by a linguist, so he obviously never said such things :)). That's a horrible misconception of course, since there is no language, as far as I can tell, that has a flat accent. Even languages that make no accent distinction (e.g. French, Polish and Standard Korean) have tonal (and sometime also stress) patterns. – Boaz Yaniv Jun 14 '11 at 19:54
  • 11
    Linguists refer to English as having a *stress accent* and to Japanese as having a *pitch accent. Another word they use in the case of Japanese is downstep*. – hippietrail Jun 14 '11 at 23:43
  • 2
    My experience in a Berlitz class was that the class changed dramatically once they handed out the books. The students correctly muted the vowels when they were just imitating the teacher aurally - but then voiced each and every one of them once they were trying to read out of a book. I found it fascinating that nobody seemed to make the connection between the words they'd been saying correctly for weeks and the words that were put in front of them. –  Jun 14 '11 at 15:59
  • 3
    I didn't see it mentioned anywhere on this page, so: this phenomenon is called 'Vowel Devoicing" https://en.wikipedia.org/wiki/Japanese_phonology#Devoicing – BlueRaja - Danny Pflughoeft Aug 19 '16 at 20:25

1 Answers1

46

I have a book in my university library that has a 100-odd page article dedicated to these mute vowels, and it still doesn't seem to give a complete picture. So unfortunately, this feature of Japanese phonology is quite complex.

Still, there's a rather simple rule of thumb that can point you to most of the places where muting may occur (and in most of them it does occur, most of the time :)). It goes like this:

  1. The vowel must be a short i or u.
  2. The consonant before the vowel must be voiceless:
    /k/, /s/ (also includes しゅ), /t/, /h/ (ふ and ひ), and maybe also /p/ (though it seems rarer).
  3. The vowel must be at the end of a word, or followed by another voiceless consonant.

This explains why you see muting in sukoshi and hikari but not in sugoi and bikkuri.

Another useful thing to remember is that you can't have two muted vowels in a row, so in words suki and tsukushita not all vowels that match rule 1-3 become mute.

Edit:

I should have given more than a passing mentions to the exceptions, because they are quite many. The rules I've given cover most of the occurrences of muted vowels, and by 'most' I don't mean 99%. It's probably not even 80%, though I'm only giving rough guesses here.

So here are some exceptions:

  • [bikkrishta] for びっくりした is quite common.
  • Sometimes (in really fast speech) some very specific grammatical forms get their vowels elided, even when the vowels are not /u/ or /i/. For instance, わからない can be shortened to [wakarnai]. It usually goes further than that with /r/ assimilating to the /n/, and thus you get the わかんない which you very often find in writing.
  • Tsuyoshi Ito and Kdanski have mentioned [sbarashii] [sgoi] in the comments.

There are of course many more. This is just an example why this issue is complex.

Boaz Yaniv
  • 12,623
  • 1
  • 50
  • 58
  • 5
    Could you provide title/author of this book? – liori Jun 14 '11 at 13:59
  • 1
    I realized that I sometimes mute the “u” in subarashii (seemingly depending on how loud I speak) although it breaks the rule 3. Probably my pronunciation is nonstandard. – Tsuyoshi Ito Jun 14 '11 at 14:14
  • I also have heard すばらしい and すごい with a muted す. Non-standard maybe, but probably not uncommon. – Kdansky Jun 14 '11 at 16:03
  • @liori: I might be wrong, but I think it's "World papers in phonetics Festschrift for Dr. Onishi Kiju". I've just took that name from my library site, because I'm almost positively sure the book that contained the article was indeed a Festschrift. – Boaz Yaniv Jun 14 '11 at 18:24
  • @Tsuyoshi: I don't know whether it's standard or not, but as I've said before, these rules have quite a few exceptions. That's why I call them rule of thumbs - they help you capture a great deal of the most common muted vowels. – Boaz Yaniv Jun 14 '11 at 18:25
  • I corrected びっくりした but am not sure whether you intended to remove the gemination from the pronounciation. –  Oct 19 '11 at 16:49
  • Excellent answer. I just wanted to add that devoicing is generally a gradual process: a voiceless consonant will always somewhat devoice the following vowel at its onset, and consequently, the shorter the vowel the less likely it is that it will be voiced, and the longer the vowel, the more likely some kind of voicing will surface. "desu" is often noted as a perfect example of devoicing, but it's not -- it's quite common to hear people say something like "sou desuuuu" and make the u long and voiced. – alexandrec Oct 25 '11 at 14:30