Re: [tied] -affa (was: The role of analogy, alliteration and sandhi

On Sun, 17 Oct 2004 12:36:46 +0200, alex
<alxmoeller@...> wrote:

>
>Richard Wordingham wrote:
>
>> Richard:
>> PIE *kW and *k^w (I take it k'W is a mistyping of k'w) have
>> different reflexes. *kW yields Centum /kw/ and Satem /k/. *k^w is
>> a cluster, and yields (generally) Centum /kw/ and Satem *k^ reflex +
>> w. The best known example of *k^w is in *ek^wos 'horse', which
>> yields Sanskrit as'va-. Another example of the cluster is in
>> *k^won 'dog', but complications arise with the zero grade.
>
>something is not clear to me in this root.Apparently the "k^w" in this root
>is proposed just because of the Avesatn/Sanskrit reflexes ("s" instead of
>"c^" ).

No. The PIE cluster *k^w yields Sanskrit <s'v>, Iranian
<sp>, Albanian <s>, Lithuanian <s^v>, Slavic <sv>, Armenian
<s^>, all different from the result of the single sound *kW,
which gives Sanskrit, Iranian, Albanian, Lithuanian, Slavic,
Armenian /k/.

>For all other languages the labiovelar should work just fine.

The centum languages generally merge *k^w and *kW, but
probably not Greek, which distinguishes *k^w > *kkW >
kk/(tt)/pp (hippos, hikkos) from *kW > Myc. <q> (= /kW/) >
later k/t/p (depending on what follows).

>If for other languages there is a normal reflex of "kW", then is not wiser
>to think the palatalisation of "c^" to "s" in Sanskrit will be a later
>reflex of a regular "c^" from an IE "kW"?

You're hopelessly confused. *k^ in Sanskrit always gives
the shibilant <s'> (not <s>), regardless of what follows (so
*k^w -> s'v). *k and *kW merge to Indo-Iranian *k, which
normally gives <k>, unless followed by a front vowel (or
ausgeglichen in a paradigm), in which case we have <c> (=
/c^/).


=======================
Miguel Carrasquer Vidal
mcv@...