30-09-03 18:30, Miguel Carrasquer wrote:
> What can we make of all of this?
I think we should restrict ourselves, in the first place, to words
meaning 'dew, moisture' or something not very different. Pokorny is
excessively inclusive here. We should therefore begin with Lat.
ro:s/ro:ris (m.), Balto-Slavic *rasa:, Old Indic rása- (m.) and rasá:-,
the Indo-Iranian hydronym *rasa:-, leaving less certain material aside.
A connection with *h1er-s- 'flow, run' is possible but somewhat
speculative, whereas words meaning 'hate, anger' etc. had better be
considered separately.
> Some forms present rather clear evidence for a laryngeal [specifically,
> *h2] in the root. We have *(h1)ró(:)h2s- (Latin ro:s, Alb. resh, Grk.
> ero:e:), *(h1)ré:h2s- (Gmc. re:s > ra:s), *(h1)r&2s- (Grk. erao:, Gmc.
> ras). The zero-grade form looks problematical (why *r&s- instead of
> *r.:s-?), which may or may not be related to the question of whether the
> root began with *r- or with *h1r-. Even if a zero grade *r&s- > *ras-
> could explain Balto-Slavic *rasa:, that option is not available, I think,
> for Sanskrit rasá:, rásas, because *&(1/2/3) in principle should have given
> /i/ in Indo-Iranian (and *r.:s- > i:rs.- as in i:rs.yati).
Agreed.
> That would seem to leave metathesis (*roh2sah2 > *rosh2ah) as the only
> possible, but unsatisfactory, explanation of the Sanskrit forms. Any
> better suggestions?
If this were indeed the only possible explanation, we'd have to treat it
as satisfactory, but I can see other possibilities. The Latin noun is a
consonantal stem, so the quantity surely requires no laryngeal
explanation. All one needs is length generalised from the original
nom.sg., as normally in Latin (<auro:ra>, <cruo:ris>, etc.). Perhaps
<ro:s> is an old collective ~ mass noun related to *h1er-s-, which would
allow us to posit an original paradigm like *h1ro:s, acc. *h1ros-m.,
gen. *h1rés-(e)s (or the like).
Now, Sanskrit has no Brugmannian length in nouns like <dáma-> 'house,
home' and <mára-> 'death', where other branches point to *o. An
otherwise invisible laryngeal suffix may often be the reason, as it
certainly is in cases where the laryngeal manifests itself in some other
way, cf. <rátha-> 'wagon, chariot' < *rót-h2-o-. This makes the analysis
of <rása-> 'juice, liquid' as *h1rós-h2-o- a not unjustifiable
possibility. At the very least, what works for <dáma-> ought to work for
<rása-> as well. I'm not sure about the best derivation of <rasá:>
'moisture'. It must be the same thing as BSl. *rasa:, so anything like
*h1res-áh2 is ruled out. *h1ros-h2-áh2 looks cumbersome but may be the
right solution (related to *h1ros-h2-o- like Lat. rota to Skt. rátha-?).
Piotr