In the _Elements of Indo-European Phonology_ published on the TITUS'
website (
http://titus.fkidg1.uni-frankfurt.de/didact/idg/idgphon.htm)
*h3 is ascribed a phonetic value of [voiced faryngeal fricative], and
it's stated there, that it therefore can trigger regressive
assimilation, switching a preceding voiceless consonant to its voiced
allophone. Thus, it's stated, that the cluster /ph3/ phonetically
surfaces as [b`] in the same manner as /pd/ surfaced as [bd]; the
examples provided include 3 pl. present of the drink-verb,
reconstructed there as *piph3enti [pib`onti], allegedly continued in
OInd. _pibanti_, Lat. _bibunt_, OIr. _ibat_ and *h2eph3o:n [hab`o:n]
'watery' (N.sg), continued in OIr. _aub_.
That raises a lot of questions.
1. Is anybody on the list aware of other examples, both pro and
contra?
2. If Piotr's treatment of the drink-verb
(
http://groups.yahoo.com/group/cybalist/message/7986, also
http://groups.yahoo.com/group/cybalist/message/7968) is correct,
*piph3enti is an impossible form, the correct form being *pipiHenti
(since 1. the root would be *peiH- rather that *peH- + *-i-; 2. the
exact quality of the laryngeal is unknown, since -o- in forms like
*po:- and *poi- is accounted for by the qualitative ablaut rather
than the o-colouring laryngeal; 3. the word looks like an athematic
intensive from athematic *peiH-/*poiH- formed by "i-grade
reduplication").
But then, I with my rather dilettanticly straightforward way of
thinking can't see how *pipiHenti would account for, say, OInd.
_pibanti_ and Lat. _bibunt_. Why _b_, indeed? Why Latin -unt- in case
the laryngeal is not restricted to *h3?
And, last not least, why *piHV- > *pV? One would expect *pijV- (if
from *pi.HV-) or at least *pjV- (if from tautosyllabic *piHV-, this
would assume the sonority of the laryngeal to be greater than that of
*i [j].
Sergei