On 2007-06-05 01:31, Sean Whalen wrote:
> The PIE form of these words can't include *dzt
> because their outcomes often show *t>0, leaving the
> other as d, not *dz.
IF all these words contained a *t in the first place. The contrastive
suffix often appears as *-ero- rather than *-tero-, also when added to
adpreps in obviously old formations, e.g. *h1ep-ero-, *sup-ero-,
*n.dH-ero-, etc.; the *-t- is probably due to reanalysis (*-t-ero- -->
*-tero-). The fate of *-d-tr/lo- in Latin can be deduced from the
behaviour of _irregularly_ formed instrumental nouns. From <ro:de:re> we
have <ro:strum>, compatible with the assumption of *-d-tr- > *-tstr- >
-str-, with the usual failure of the second *t to assibilate before an
*r. Then we have <caelum> 'chisel' from <caedo:> 'cut', <sca:lae>
'ladder' from <scando:> 'climb', and a few similar forms, suggesting
something like *-d-tlom > *-s(s)lom > -(:)lum.
<ra:dula> and <pendulum> are, in my opinion, new formations, parallel to
<te:gula, re:gula> from <tego:, rego:>. In old instrumental nouns the
variant *-tro- was preferred after roots containing a liquid, and this
is what we find in <ra:strum> (also from <ra:do:>!), but rather than
preserve the regular reflexes of *-tr/lo- formations (no longer
transparently relatable to their verbs), Latin developed the innovated
suffix -ulum/-ula as an allomorph of -culum/-cula/-crum/-cra and
-bulum/-bula/-brum/-bra after roots with a final stop. This is an inner
Latin affair and tells us nothing about the PIE treatment of *-T-tr/lo-
(surviving only in traces in Latin).
Piotr