On Thu, 22 Jan 2004 16:26:25 +0000, elmeras2000 <
jer@...> wrote:
>--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
>[...]
>> Do you mean something like:
>>
>> **ka -> *k
>> **ki -> *k^
>> **ku -> *kW
>
>Yes.
>
>> The problem with such a hypothesis is that then we would expect the
>> frequency of *k^ to be about the same as the frequency of *kW,
>when in fact
>> it's much higher. It is in fact the frequency of *k which is
>comparable to
>> that of *kW, both well below the frequency of unmarked *k^.
>
>No, we should expect the frequencies of the three velar series to
>match that of the earlier vowels as they were before they coalesced.
>How frequent each one was - and how many there were of them - will
>probably remain open for some time. The IE velar system may even be
>the best piece of evidence for now.
>
>Incidentally, if the same thing happened to IE itself (and in Indo-
>Iranian one part of it did), then the common weight of the
>palatalizing vowels /i/ and /e/ would completely outbalance that of
>the potentially rounding /u/ and /o/. So, all it takes is that there
>once was a language much like Indo-European. I could imagine that.
It can be imagined, but it's not the most likely possibility. If we're
talking about a 3-vowel system a/i/u, /a/ is likely to be more frequent
than either /i/ or /u/. In a 5-vowel system (a/e/i/o/u), it's possible
that /e/ and /i/ together outnumber /a/ (so *k^ is more frequent than *k),
but it's again very unlikely that /o/ and /u/ together would have such a
low frequency (*k^ is much more frequent than *kW).
The PIE vowel system (a/e/o, with /a/ very rare) is an unlikely (but real)
one. It surely developed out of an earlier system by the merger of several
vowels into one (/e/). My personal theory is that earlier /a/, /i/, /u/
merged to /&/ (> /e/), with /o/ the result of the merger of long /a:/ and
/u:/ > /o:/ > /o/ (/i:/ gave /e:/). I cannot and do not believe that it
developed out of a system that was as unbalanced as that of PIE itself. I
also cannot and do not believe that centum /k/, /g/, /gh/ are likely to
derive from palatalized segments. My conclusions are therefore:
There were two series of dorals stops: velar /k/ etc. and uvular /q/ etc..
with the initial frequency of /k/ etc. naturally higher than that of /q/
etc. In the neighbourhood of **u(:), /k/ and /q/ developed into /kW/ and
/qW/. Palatalization in the neighbourhood of *i(:) may have occurred as
well, but was never phonologized. I would imagine something like:
*ka > *k^e *qa > *ka
*ki > *k^e *qi > *ke
*ku > *kWe *qu > *kWa
*ka: > *k^o *qa: > *ko
*ki: > *k^e: *qi: > *ke:
*ku: > *kWo *qu: > *kWo
If we assign theoretical weights to the frequencies (e.g. /k/ 3 times as
frequent as /q/, weight of /a/, /i/, /u/, /a:/, /i:/, /u:/ something like
4, 3, 3, 2, 1, 1), the PIE distribution comes out as:
*k^ 12+9+6+3 = 30
*kW 9+3+3+1 = 16
*k 4+3+2+1 = 10,
which looks about right.
=======================
Miguel Carrasquer Vidal
mcv@...