On Sat, 02 Feb 2002 21:06:08 -0000, "michael_donne"
<
michael_donne@...> wrote:
>> A development k > s is much more common than a development s > k, so
>> *k(^) would be the preferred reconstruction even if there were no
>> other evidence available (which there is, but it's almost six in the
>> morning here).
>
>I'd be very interested in hearing more about the other evidence --
>after you've gotten more sleep! :-)
I had a birthday party to organize and attend to, so I couldn't reply
earlier. And now that I do reply, it's almost six in the morning
again... But I can't sleep anyway.
Off the top of my head: centum *k^/satem *s' rather often shows
irregular developments (namely /k/) now in this, now in the other
branch of the satem group (on the whole more often in Balto-Slavic and
less in Indo-Iranian). Slavic <slovo> "word", but Lithuanian
<klausyti> "to hear" (from PIE *k^leu-/*k^lou-). Sanskrit <s'mas'ru>
(for *<smas'ru>) "beard", but Lith. <smakras> and Arm. <mawruk`> (<
*smakrues) (from PIE *smak^ru-). In the centum languages *k^ is
always /k/, never something ess-y (except where it later became
palatalized before front vowels). Now if we have *k in one group, and
either *s' or *k in the other, it's reasonable to suppose that the
oldest stage is represented by *k (or *k^).
And of course we can't look at *k^ in isolation. *k^ is merely a
single element within a larger phonological system. Where Sanskrit
has /s'/, /j/, /h/ (Av. /s/, /z/, /z/; Arm. /s/, /ts/, /dz/
(transcribed <s>, <c>, <j>), etc.), the centum languages correspond
with /k/, /g/, /gh/. All of the stops in PIE come in these three
varieties: unvoiced/unaspirated, voiced/aspirated and
voiced/unaspirated, and none of the PIE fricatives come in them (if
it even makes phonetical sense to talk about aspirated fricatives).
This must mean that the sounds in question were originally (velar or
palatal) *stops*, never fricatives. One may choose to notate *c, *J
[I mean IPA j-bar], *Jh to stress their palatality, but another,
equivalent (though not IPA compliant), way to write that down is
precisely *k^, *g^, *g^h.
>> How does Beekes question Hittite?
>
>In "Comparative IE Linguistics", page 31, he says:
>"... Hittite makes a very archaic impression owing to its simplicity,
>and its verb could represent a system older than that of Sanskrit or
>Greek. This is the reason why it was thought that Hittite must have
>split off much earlier from the other languages. This is called the
>Indo-Hittite hypothesis.... This would mean that the other IE
>languages underwent a common development which Hittite did not
>undergo -- otherwise the the hypothesis has no import. I am not yet
>convinced by it: Hittite is fraught with too many fundamental
>problems to justify such a far-reaching conclusion as this one....
>Hittite has lost much of its oldest data (probably under the
>influence of people who originally spoke another lanuage), so that
>its simplicity need not suggest antiquity, but could rather be owing
>to loss."
This doesn't question Hittite or its antiquity, it questions a
particular view on the consequences one should draw from the Hittite
data. Basically, there are two options (although the truth is
somewhere in the middle): (1) Hittite is not only the oldest attested
IE lg., it's also the language that stands closer to how PIE really
was. In fact we'll rename Indo-European to Indo-Hittite to stress
that point. Indo-Hittite spawned two daughter languages: Anatolian
(Hittite), which did not innovate much; and Indo-European (all the
rest), which developed a bunch of innovative features (feminine
gender, aorists and such) and dropped a set of others (stative
hi-conjugation, etc.). (2) Hittite, while being the oldest attested
IE lg., is in fact one of the more innovative ones, and particularly
prone to lose features (feminine gender, aorists, etc.), while it also
developed a few unique things of its own (stative hi-conjugation,
etc.). Let's keep the name "Indo-European."
>If Hittite is not as old as thought, that would make Indo-Iranian
>perhaps the oldest language (although I'm not suggesting it is PIE or
>anything.) Since IE studies are commonly admitted to have focused
>more on European (i.e. Centum) languages
Not true. Sanskrit has always been at the focus of IE studies.
>and since most IE languages
>are Centum (Nichols explains how this is predicted by their being at
>the periphery of the spread zone),
But the periphery maintains the archaisms. There is no possible
connection between all the different centum languages, while the satem
languages form a geographically connected group, which shares a number
of other features besides their treatment of the palato-velars (e.g.
the development {r,u,k,i}s > s^). So either all the centum languages
have independently happened to shift their *s''s to *k's, or the
centum languages (there is no "centum group") preserve the original
situation, and the satem group has innovated. You don't need a degree
in probability theory to guess at the odds.
=======================
Miguel Carrasquer Vidal
mcv@...