On 2007-09-09 08:08, stlatos wrote:
> *drusdo+ > *druzdo+ / *druzgo+ > (various) 'thrush'
I've often wondered about this one myself. The problem with this word is
its extreme variability, which makes even the reconstruction of the PIE
prototype somewhat doubtful. *dru-sd-o- is my own proposal, so I'm not
going to argue against it, but no matter if the original shape was that,
or *trozdo-, or indeed anything, the various reflexes cannot be
explained without allowing rather liberally for distant assimilations,
dissimilations, folk-etymological distortions and other special effects.
Even *drusgo- has to be further manipulated to give OE þrysce.
> *doru 'wood, tree'
> *drustxo+ 'standing _ = tree' > Alb drushk 'oak'
>
> *xYrurdhto+ > OE ru:st; Alb ndryshk
>
> *xYrurdhto+ > L ruscum 'butcher's broom (plant w red berries) >>
> rusco: / rusto:
>
> *pr,xWisto+ 'first' > L pri:sti:nus, pri:scus
>
> *proxW stx,to+ 'set (straight) forward' > OCS prostU;
I'd analyse this one as *pr[h3]o-sth2-o-, with the *h3 lost in
composition so early that *pro- counts as PIE.
> Khow frosk
> 'straight'
>
> *lals+ > Skt las.-
> *lalsti:vus > L lasci:vus
Don't you feel Slavic *lasU 'covetous' and *laska 'grace, love' should
belong here? In Skt. las.ati, where reduplication might be expected, I
have little objection to *lV-ls-, but why in Latin? The root seems to be
*la(:)s-, perhaps *lah2s- (esp. if Schrijver is right about the
development of *R.hC- > Lat. RaC- in absolute anlaut). Gmc. *lustu-,
*lusti- looks aberrant, but the initial zero-grade of *(C)RVC- roots
(expected *(C)uRC-) may be "corrected" to *(C)RuC- in Germanic. So you
could have the abstract noun *l.h2s-tu/i- vs. adj. *l.h2s-k(^)o-
(underlying Lat. lasci:vus).
> *xWostn.+ 'a bone' > Arm. oskr
Compare with this the velar extension of Skt. asrg/k, gen. asnás 'blood'
(h1esh2-r/n-) and forms like Gk. óstrakon < *h2ost-r.-k-o- and Welsh
asgwrn. My gut feeling is that the velar originally followed the
heteroclitic *-r/n- and was metathesised in Armenian and Brittonic
*h2óst-r.k > *h2ós(t)-kr. . But the relevant forms are so few that any
conclusions drawn from them are doomed to be tentative. If Lat.
sanguen/sanguis reflects *h1sh2an-gW-, that casts doubt on the
assumption of one and the same velar increment in 'blood' and 'bone'.
I've no idea how to analyse them. *h2óstr.k might represent *h2óstr.t
Cf. yakr.t) with place dissimilation (and if so, we might have
*h2oskr.(t) as a result of the same dissimilation working in regressive
mode). The extended form of the 'blood' word looks complex enough to
reflect an obscured compound mixed up with the normal paradigm of its
first member (something like Iuppiter : Iovis).
> These words and others are part of something I believe was
> widespread and involved in early IE separation; I've mentioned some
> before and can gather the parts together if needed, but I'd like to
> know if you have a different explanation.
The situation is complicated by the fact that there are many derivatives
in *-k^o-, *-(i)sk(^)o-, *-(i)ko-, *-h3kW-o-, *-gW[h2]-o-, which may be
added to the same bases that form derivatives in *-to/u/i-, *-isto-,
*-st[h2]-o- and the like, so a fortuitous correlation can be mistaken
for a phonological correspondence. What would you make of
*h2ju-h3n.-táh2 'youth' (Lat. iuventa, Goth. junda) vs. *h2ju-h3n-k^ó-
(Skt, yuvas'á-, Gms. junGa-)? If this alternation is morphological
rather than phonological, why should not the same be true of Goth.
lustus : lat. lasci:vus?
Piotr