On 2006-05-06 09:47, Piotr Gasiorowski wrote:
> --- In cybalist@yahoogroups.com, Andrew Jarrette <anjarrette@...>
> wrote:
>
>> Yes, but I have a further question: did the same devoicing occur in
>> OHG, to account for its genitive singular (a-stems) in -es? And
>> why does it have -e- if this ending is truly from *-oso?
>
> The OHG ending was borrowed from pronouns such as <des> (from
> *tes(j)o, with accented *e).
I forgot to answer the first question. I find the OHG forms somewhat
problematic. The -es ending of the gen.sg. doesn't decide the issue
either way. The original nom.pl. of a-stems was lost in pre-OHG times.
In the verb system, forms such as 3sg. -it, 3pl. -ant (instead of *-id,
*-and) militate against devoicing, while 2sg. -is supports it. Actually,
it's _only_ this particular ending that suggests devoicing, so perhaps
some special explanation applies here. One rather obvious possibility is
that the combination of 2sg. -z(i) with enclitic *þu(:) produced *-s-tu
> *-st earlier than might be inferred from the testimony of OHG texts,
and that -s is a back-formation (*-s-tu minus the enclitic).
If there's any truth in this explanation, the devoicing of final
fricatives was dialectal ("North Sea") within WGmc., while OHG preserved
the primitive state of affairs. The spread of z-rhotacism (applicable,
of course, only to the synchronically surviving instances of */z/) took
place in an already diversified WGmc., but that's compatible with other
evidence.
Piotr