In this Japanese song, there is a line saying:


The /d/ in 「で」 and the /b/ in 「ぼ」 sound to me as if they (especially the /b/) are pronounced as unaspirated /t/ and /p/ respectively.

Some possible explanations:

  1. I simply heard them wrong.
  2. The singer's pronunciation here is kind of unusual.
  3. People sometimes use unaspirated voiceless sounds instead of their voiced counterparts, even at non-word-initial positions (an answer saying people do this at word-initial).

Please kindly share your opinions!

added a clearer clip of the line (source)

Somehow, the /d/ in this clip sounds more unvoiced to me than the /d/ in the youtube video above <@_@>

  • 1
    My by ear assessment is the same as snailboat. These both sound very clearly to me as /d/ and /b/. Snailboat and I are both native speakers of AmE for what it's worth.
    – virmaior
    Commented Sep 12, 2015 at 19:03

2 Answers 2


I think it's half #1, half #2.

When a singer (especially female) sings tenderly, some whispering (breathy) feature tends to be blended into the voice, resulting in incomplete voicing. It's a universal phenomenon. The whole phrase in your sound clip has underlying breathing, so in some ways you're true, these are not true voiced consonants.

What you hear at で or ぼ is what they call slack voice. Your vocal cords are set in the position where you pronounce voiced consonants, but the vibration is less than perfect, making a sound half voiced, half unvoiced. If you only focus on VOT (or, if there is vibration during plosion), you may think it's unvoiced, but you must also care about the sound quality, so that you'll notice it's different (much "thicker") than ordinary unvoiced consonant, namely か in your clip.

What's more, the singer properly tries to maintain glottal vibration till right before those consonants, which makes the consonants of で and ぼ effectively sound like //dt// and //bp//, unlike か's is just //kʰ//. But when it comes to this point, で's "voicedness" is certainly quite weaker than ぼ, as you observed.

cons.    voice offset    voice onset
 で          80ms          50-70ms?
 か         200ms           100ms
 ぼ          10ms            10ms

(shoddily analyzed with Praat)

  • This post is a lovely example of why it's great to have people who know what they're talking about around. +1! Commented Sep 13, 2015 at 7:33

I think it's #1, you heard them wrong. They sound like /d/ and /b/ to me.

I'm a non-native speaker, but the /d/ and /b/ don't sound unusual to me.

  • Thanks! I've just added a clip that sounds clearer to me to the question. Could you please check out the clip as well? :)
    – null
    Commented Sep 12, 2015 at 18:05
  • 2
    @Noir Thank you for the clip! It does sound clearer. But I still can't hear it as /t/ and /p/ no matter how many times I listen. Unfortunately, because the music is present I can't use a tool like Praat to show that it's voiced, so all I can do is tell you what I hear, and I hear /d/ and /b/.
    – user1478
    Commented Sep 12, 2015 at 18:10
  • Thanks for the feedback! It's very likely that I am confusing voicing with something else :|
    – null
    Commented Sep 12, 2015 at 18:25
  • 3
    @Noir I see from your profile that your native language is Chinese. Chinese doesn't have voiced stops, so it's common for native Chinese speakers to hear G, D, and B as K, T, and P respectively. You'll just have to train your ear.
    – Blavius
    Commented Sep 13, 2015 at 4:27
  • 2
    @Noir You can use Praat, sure. You can probably find a tutorial online. (As far as I can tell, though, Praat is useless for this sound clip.)
    – user1478
    Commented Sep 13, 2015 at 20:46

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .