final devoicing in American English?

[Mark_Mandel at DRAGONSYS.COM]

>From the two A's, Alice and Andrea, we have the following

Alice Faber wrote:
>         Something
> came up in class today that I haven't previously encountered. I was doing
> the standard presentation of regular English plurals, so that the students
> could discover the phonological conditioning of /-s/, /-z/, and /-Iz/
> variants. To my surprise, several of the students insisted that /z/ in
> _dogs_ is at least partially voiceless. They weren't getting tricked by the
> orthography. (Many of them were quite positive that _tree_ begins with an
> affricate, for instance, and they were making fairly subtle vowel quality
> distinctions.) Two of them, one from northern Georgia and one whose first
> language is Russian (though her English sounds virtually accent-free), had
> more conventional final devoicing, including regressive assimilation, so
> that _dogs_ ends /ks/. These are all Wesleyan University undergrads of
> normal college age.

<<<Andrea Vine replied:>>>

Funny you should ask this.  When I was an undergraduate (this would have been
about 18 years ago), these very issues came up in an intro ling class.  One
student insisted that "tray" began with an affricate (and it did, the way she
pronounced it) and another pronounced "bugs" with a final /s/, not /z/.  The
"bugs" student was from California - I don't know where the other student was
from.  But they were both Asian-American, sounding as though they had been born
here.  I remember wondering if there was any Asian influence, by way of
community or some such, to cause the particular sound patterns.

<<<<<
First the affricate. I remember as a grad. student at Berkeley insisting to Mary
Haas that my "tree" begins with an affricate; she didn't hear it that way. Now,
the standard AmE aspiration of the initial /t/ makes the /r/ at least partly
voiceless. In my case, my /r/ in most positions is "humped" (middle of tongue
raised, convex upward) rather than retroflex (tip raised, concave upward). I was
very surprised on first reading, I forget where, that this is rare among AmE
speakers, who mostly retroflex, and that the acoustics are identical. But in
this context my /t/ assimilates in tongue position to the humped /r/ and is
palato-alveolar, not as far back as /C/ ("ch") but well back of my prevocalic
/t/. As a result, the tongue contact with the upper surface of the mouth is much
deeper, front to back, than with a prevocalic /t/, and the release of voiceless
breath during the aspiration creates friction that I can feel and hear. Thus, an
affricate. (BTW, my background: USA, male, born 1948, grew up in Westchester
County north of NYC to age 7, then in NYC. Parents both NYC born, college
degrees.)

The (partially) voiceless final sibilant in "dogs". I've often noticed a similar
phenomenon in Garrison Keillor's pronunciation on NPR; he's from Minnesota.
Alice's Russian-native speaker might also be devoicing root-final obstruents as
Russian does and assimilating progressively.

-- Mark A. Mandel