Re: [tied] Olsen's Law [was: PIE Ploughs]

On Fri, 23 Jan 2004 16:36:45 +0100 (MET), Jens Elmegaard Rasmussen
<jer@...> wrote:

>On Fri, 23 Jan 2004, Miguel Carrasquer wrote:
>
>> According to Jens, the root *poh3- is actually *poh3i- (with possible
>> contextual variants *poh3-, *poih3-, *pih3-, *ph3i-), so I was wondering
>> whether the possibility has been investigated that the different
>> treatment
>> depends not (only?) on the quality of the laryngeal, but (also?) on the
>> presence/absence of this *i in the root.  Just a thought.
>
>It has, and the semivowel has no part in it. Still, I have assumed
>preaspiration even before lost /H2w/ in IE *térthro-m, Gk. térthrom
>'summit' from older *térH2w-tro-m. But that also just skips the semivowel.

OK.

>> >I'm not sure whose idea it was originally. Something of the kind was
>> >certainly advocated by Kurylowicz (that's probably where Watkins had
>> >it from). The problem is that we don't find *dH where expected -- but
>> >then, Olsen's Law can be interpreted as a PIE precursor of
>> >Bartholomae's Law. It's very much the same thing -- aspiration by
>> >progressive assimilation.
>>
>> I was thinking whether there was a way to combine Bartholomae's law (with
>> its effect /t/ > /dh/) with Olsen's preaspiration law to explain the
>> result
>> /d/ in Slavic. 
>
>I may be biased, but why does this have to be assigned to Bartholomae if
>he did not write about it? The Slavic -dl- is not restricted to aspirating
>contexts.

Slavic has generalized -dl-. The question is why there was -dl- in the
first place. Bartholomae's law can be written as Xht > Xdh, so all we have
to do to explain the Slavic phenomenon is widen the scope of X, so that it
can be any consonant (followed by breathing) whatosever, including zero.
If the law worked in pre-Slavic, that would have given roots ending in a
voiced aspirate [Bartholomae proper] and roots ending in a laryngeal
[Olsen] an ending *-dhlo-, which then spread to all contexts. As with
Bartholomae's law in its traditional formulation, the problem is one of
scope and/or timing. Did it apply to PIE as a whole, or was it a pre-Vedic
(c.q. pre-Slavic) thing only? Was the working of the law undone by later
analogy or was its scope limited (in which contexts?) to begin with? I
have no other cases to offer in Slavic where we find /d/ for expected /t/
(except <gospodI>, but that doesn't seem relevant). There are the two
Armenian words with unexplained *dh < *th, which Birgit mentions (awd and
awdik`), and, if I'm right about *dh > r in Armenian (2pl. aor. pass.
-aruk` < *-a:-dhwes), then also the 3sg. imperfect (< optative) ending -yr
< *-yeth < *-yeh1-t (paradigm: "I was" eí < *(e)s-yéh1-m, eír <
*(e)s-yéh1-s, éyr < *(e)s-yéth).


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