Re: half voiced (was: Digest Deux)

Someone wrote:
>       I'll venture a short explanation to draw other more learned
> citizens out.
>       Cosonants are usually voiced intervocalically and
> word-finally.  This is only "usually", however, since they can appear
> voiceless.  The main key in telling the difference between voiced and
> voiceless stops in English is aspiration.  Intervocalically and in
> non-stressed positions is the only place where you're likely to find some
> variation, and even then, some strings have been pulled, so that /t/ and
> /d/ are always [4] (alveolar flap), so that there is no distinction.  /g/
> is never voiced, apparently--only very rarely, and that has to do with
> the aerodynamic voicing constraint.  So the two velar stops in English
> are [k] and [k_h]. When the issue of "half-voicing" comes up, what that
> means is that is that the sound starts out voiced, but ends up being
> voiceless, and this also has to do with the aerodynamic voicing
> constraint.  Say in the word "bad".  You'd expect something like [b&:d],
> but you end up with something more like [p&dt], where the consonant is
> initially voiced, but (and this happens especially utterance finally)
> when you come to the end, you tend to let the voicing go, and the end of
> the [d] ends up being more like a [t].  You can verify this with sound
> analysis software, such as PRAAT (the one I use).
Actually the <d> in "bad" is not so much "half-voiced" as it is
weakly articulated. Its important to remember that another important
perceptual cue is _articulatory strength_. English /p t k/ are fortis,
while /b d g/ are lenis. This is most evident in syllable-final
position. In some dialects, both series of stops are completely voiceless
in syllable-final position. In these dialects, the perceptual cue lies in
the length of the preceding sound. E.g. /b&d/ "bad" vs /b&t/ "bat" is
[b&:t] vs [b&t] respectively (where [&:] is suppose to be half-length,
not full-length). Another example is /&nd/ "and" vs /&nt/ "ant", which is
[&n:t] vs [&nt] respectively (where [n:] is again half-length). So with
regards to "half-voicing", the part of the final /d/ that sounds like is
voiced is actually the slightly lengthened sound before /d/, while the
part that sounds voiceless is actually the /d/ itself. Syllable-final /d/
in effect may therefore sound "half-voiced" to the untrained ear.

Just thought I'd mention that since no one else has.

-kristian- 8)