--- In
cybalist@yahoogroups.com, Andrew Jarrette <anjarrette@...> wrote:
>
> Does anyone know the reason why Sanskrit has <h-> in <hrd> "heart"
(also <ha:rdi>) where non-Indo-Iranian languages all have evidence of
*k^-? I think also Avestan has <z-> in this word too.
> Similarly, why does <hanu> "jaw" have <h-> where non-Indo-Iranian
IE all show *g^-? I remember the theory that <dva:r> "door" has <d->
instead of <dh-> by the influence of <dva:> "two" (because doors often
come in pairs, double-doors), but what could be the explanation for
<hanu> and <hrd>?
>
The second can easily be explained by reconstructing the root as
*g^h1enu. In all languages except IIr. the laryngeal is lost. In IIr.
it's lost as well, but with aspiration of the preceding stop. This
also explains why IIr. shows aspiration in the first person nominative
singular pronoun whereas the other languages point to a plain voiced stop.
The first example is harder to understand. Laryngeals can't help us
there unless we posit an h4 that voices and aspirates preceding stops.
A couple weeks ago I did come up with a way of reducing the PIE system
of stops from a 3 way contrast to a 2 way contrast. I propose that
stops were either voiceless aspirates or voiceless ejectives. There
are two basic rules for recovering the traditional reconstruction and
two rules of allophony.
Basic rules:
1) Ejectives become plain voiced
2) Aspirates become voiced aspirates
Allophony:
1) If a root contains two ejectives the second dissimilates to a plain
voiceless stop
2) If a root begins with a cluster of glottal stop plus stop, then any
aspirates in the root lose aspiration (and therefore become plain
voiceless stops).
If we assume the existence of a glottal stop prefix then certain
alternations become perfectly understandable. In the above example
most languages would reflect *?k^Hert' > *k^erd but IIr. would lack
the glottal prefix and have *k^Hert' > *g^Hert'.
This can also explain:
1) kap ~ gHabH
2) ap ~ abH
3) ost ~ kost
4) ag^ ~ kag^
For (3) and (4) I assume that laryngeal hardens to a velar when
preceded by a glottal stop. An alternate explanation of (2) is that
the variation is ap ~ ab which is explained by an h3 root
determinative. This would make more sense if laryngeal hardening is true.
The above theory explains the constraints on roots (e.g. why roots
with two plain voiced stops never occur) and otherwise hard to
understand root variations. The main drawback is that you have to
assume a prefix with no apparent semantic value.