On 2007-06-13 00:44, Piotr Gasiorowski wrote:
> On 2007-06-12 20:39, stlatos wrote:
>> With your description now specifically allowing -lum and -ulum, do
>> you still think or need to have tt>ss as an explanation of sca:lo+ and
>> caelo+? What stage would you use to explain my *pangtlo+ > *pangslo+
>> pa:la 'spade, shovel'? Which stage allows the Lachmann lengthening
>> you posited before?
> This is the oldest layer, i.e. the regular development of *-K-tlo-, in
> which the *t was for some reason treated as in the "thorny" clusters
> resulting from the metathesis of *-tk-. This treatment is old and not
> restricted to Latin. More about that tomorrow.
For <pa:la> as well as <pa:lum ~ -s> 'stake, I propose *pag(^)-tlo- >
*pakþlo > *pakslo- > *pa(g)zlo- > pa:lo-, but note the dim. paxillus
'peg' from *-kslelo- > *-ksl.lo- > -ksillo-. The formation of a
secondary syllabic *l. in such derivatives is younger that the pre-Lt.
change of *-tlo- > *-klo- (see *po:tlelo- > *po:klelo- > *po:kl.lo- >
po:cillum), but older than the loss of *-(k)s- before *l.
I'm making no assumptions about the phonetic interpretation of *þ in the
intermediate reconstruction above: it's just the same treatment as in
the well-known "thorny" words, possibly with a [ts]-type affricate
rather than a special fricative. Treat my *þ a a conventional cover
symbol for the various possibilities discussed in the literature. We can
discuss it in a separate thread to avoid side-tracking.
Some other examples:
*weg^H-tlom (*'implement for propulsion') > *wek^þlo- > *wekslo- >
ve:lum 'sail' (vexillum 'flag'); cf. Sl. *veslo 'oar'.
*h2ag^-tlah2 (*'lever' >) 'upper arm, shoulder' > *h2ak^þlah2 > Lat.
a:la 'wing' (axilla 'little wing, armpit'); cf. Gmc. *axslo:. Compare
the similar development in *h2ak^-tlah2 'whetstone' > Sl. *osla.
*te(t)k^-tlah2 'carpenter's tool' > *tek^þlo- > Sl. *tesla 'adze', Gmc.
*weik^-tlah2 'dwelling-place' > *weik^þlah2 > Lat. vi:lla (note that *l
is at least variably geminated after an _etymologically long_ vowel or
diphthong -- see the next item as well as <mi:lla>, <paullus>, etc.),
perhaps because compensatory lengthening was ruled out in such cases.
*h2auk(W)-tla2 'pot' > *h2auk(W)þlah2 > Lat. aulla ~ olla (auxilla).
There is no t-spirantisation in *-K-tro-, cf. *h2m(e)lg^-trah2
'milk-pail' > Lat. mulctra, OHG kuo-melhtra*.
It's hard to say what happened to *-P-tlo- in Latin, as no good examples
seem to have been preserved.