On Wed, 04 Dec 2002 16:25:49 -0000, "Richard Wordingham"
<
richard.wordingham@...> wrote:
>gLeN's theory is that *t~*d~*dH was realised as [t]~[t:]~[d]. Thus,
>the other feature was voicing.
But the other feature is extremely *unlikely* to be voicing.
Ladefoged & Maddieson, when discussing Strength (fortis/lenis) say:
"Only a relatively small handful of languages have been proposed as
possibly having articulatory strength differences that are independent
of voicing [... only some Dagestanian languages ... Archi, for
instance, is described as having /t:/ = voiced [dd], /s:/ = voiceless
[ss] and /c:/ = ejective [c?c?] ... L & M conclude that the primary
feature here is length not strength ...]".
For PIE, it makes sense to think the voiced/voiceless contrast (*d,
*dh vs. *t) was at some earlier time a lenis/fortis contrast (*t, *th
vs. *t:), simply because voiceless aspirates are much more common than
murmured sounds (and, if the glottalic theory holds, voiceless
ejectives are much more common than voiced ejectives or implosives),
and also because two of what are probably the most archaic IE
languages, Hittite and Tocharian, do not have voiced stops at all.
But for all practical purposes, you cannot have fortis/lenis *and*
voiceless/voiced, because for all practical purposes they are
equivalent. And of the three series that PIE had, the *p/*t/*k series
is the only one that can lay claim to the fortis label (they're even
written like that, <pp>/<tt>/<kk>, in Hittite).
>Interestingly, it's consistent with my post-vocalic variation of
>Miguel's glottalisation theory (in 'Crows and the Glottalic Theory').
>With gLeN's phonetics, the first stage is reminiscent of consonant
>gemination in Italian. I'm not sure how feasible a merger of [d:] and
>[t:] would be, though.
Italian of course has both voiced and voiceless geminates (the primary
feature being length, not strength). Whether [d:] and [t:] are likely
to merge kind of depends on the rest of the system, I think. In a
system like the Italian or Old Irish one (with /t/, /d/, /tt/, /dd/),
we might expect (between vowels) a development /tt/ > /t/, /dd/ > /d/,
/t/ > /T/, /d/ > /D/. Brythonic and most of Western Romance had a
three way system /tt/, /t/, /d/, which was resolved as: /tt/ > /t/,
/t/ > /d/, /d/ > /D/, also pretty much as expected. I don't see any
natural way a system /tt/, /t/, /d/ might develop into /d/, /t/, /dh/.
=======================
Miguel Carrasquer Vidal
mcv@...