2012/5/1, Anders <
ollga_loudec@...>:
>
> Do we have corroborating evidence for this chronology, i.e. *o: > *a: before
> Osthoff's Law in Celtic? Intuitively, I'd suspect *o:RC > *oRC in Celtic,
> but I'm not aware of any examples at the moment.
Bhrihskwobhloukstroy:
OIr. alt -o-, n. 'joint, articulation, state' : Gk. péplos,
laryngealless (does it exist?) 3. √*pel- (Pokorny 802-803, Mallory -
Adams 1997: 63); Matasović's *pol-to- (121) implies a loan from
Germanic, but *po:l-to-m would be regular
OIr. bairt -i-, f. 'maiden' : Gothic barn 'child', √*bher-; the only
alternative to appertinentive Vrddhi *bho:r-ti-s would from seṭ
variant *bherH- with zero-grade and, pace Jens, derivative -h2 (from
feminine) *bhrH-th2-i-s (> Celtic *barti-s like *bardo-s <
*gwrH-dhh1-o-s with 'Schrijver's Law')
> Anders:
> Are we at all sure that OIr. áth derives from *h1yah2tu-? I'd much prefer
> linking it with MW adwy 'gap', Bret. ode, oade < *adui 'gap (in a hedge,
> etc.)', presumably a derivative *a:towyo- 'a pass, gap' from the u-stem
> *a:tu- found in OIr. áth 'ford'. Early pretonic shortening of *a: > *a in
> British Celtic is not without parallel.
>
> This would rule out a reconstruction with initial *(H)y-, since this would
> be preserved as such in British Celtic.
>
> Anders
>
>
Bhrihskwobhloukstroy:
Wonderful etymology. So *a:tow-yo-implies *a:tu-s and this could
equally explain -ate-place-names but without any particular
consequence for chronology.
1) IE etymology for *a:tow-yo-:
"origine... controversée" (Deshayes 2003: 544, for breton ode);
"no etymology" (Schrijver 1995: 217)
Semantics: adwy "1. gap (in fort, mountain, rank, page, time,
&c.); lair, nook; gap in wall or hedge; wattled hurdle, movable fence
in gap; 2. loss; crisis" (GPC I 29-30), therefore idea of void from
crash, destroying or falling down
Possible roots: *h2at- 'stick' (LIV2 274); *peth1- 'fall' (LIV2 477-478)
2) Origin of OIr. áth: if connected to adwy, the Ir. meaning
'ford' would be better explained from √*h2at- 'go' (LIV2 273-274) than
from √*h2at- 'stick' or √*peth1- 'fall'; otherwise the derivate
*a:towyo- should retain the primitive semantics and *a:tu- a
development 'stick / fall' > ' hole' < 'passage'.
On the other side, there's always the comparison with *h1yah2-
'go' (Lat. Ia:nus, partially OIr. 2án *'sun'? [O'Rahilly]), OIndic
ya:tú- m. 'one who goes, a traveller; going against, attack; sorcery,
witchcraft; a kind of evil spirit, fiend, demon; wind; time', n.
'anything to be guarded againstor warded off, harm, injury, damage; an
evil being or demon' (MW 845,2, cf. 860,1)
British *a:towyo- implies a Proto-Celtic *a:tu-; OInd ya:tu- and
the widespread distribution of *h1yah2- imply a PIE *h1yah2-tu-.
PC *a:tu- is probably from PIE *h2o:t-u- (√*h2at- 'stick') and/or
*po:th1-u- in the meaning 'gap', maybe from *h2o:t-u- (√*h2at- 'go')
in the meaning 'pass' > 'ford'; possibly OIr. áth has a double or
triple origin
But in such case why not also PIE *h1yah2-tu- 'going' > 'passage,
ford'? Even OIr. 2án is probably (Vendryes DEIA s.u.) conflation of
*h2a:s-no-, *h2ag'-no-, *h1ah2-no- etc.
A comparison with adwy is granted by internal Celtic solidarity
and by semantic closeness; a comparison with ya:tú- is granted by
internal IE solidarity and by semantic closeness as well. There's
enough room for multiple origin
3) -ate-place-names: *h2o:t-u- / *po:th1-u- or *h1yah2-tu-?
If we accept their Celtic etymologies (i.e. comparison with OIr.
áth, of whatever origin this may be), we note that non-final -u-stems
lose labialization and show palatalization (e.g. Mozzate < *Muccia:tus
with first element *mukku- 'pig', but not *mukku-(i)a:tu-s whence
†*Mocc(ogg)iate) i.e. - pace Schumacher (or at least lmited to
Continental / Cisalpine Celtic) - pi:us-Law: *muknú-HiaHtu-s >
*mukku:ja:tu-s > *mukki:ja:tu-s > Latin *Mucci[i]a:tus > Mozzate).
Therefore, a coherent treatment of the -ate layer recommends to
opt for *h1ah2-tu-s as protoform for these stems, although OIr. áth
may probably represent a conflation of etyma. Any etymon of áth would
work for the majority of -ate-names as well; only *h1ah2-tu-s works
for -ate-compounds with non-final -u-stems.
Can a sound law be limited to only a part of Celtic? This would be
a problem only if one thinks that Celtic languages have spread from a
small single point of depart; in any other model (e.g. Celticity as
result of the spread of isoglosses upon PIE dialects), there's no
difficulty (pius-Law would be a non-pan-Celtic law embracing Italic on
one side and Continental or just Cisalpine Celtic on the other side)