On 2007-03-15 01:02, Miguel Carrasquer Vidal wrote:
> Well, that was my question: the Greek forms behave as if
> derived from H.RH.(C) instead of HR.H(C). Can this be fixed
> by assuming an inner-Greek "syllabification reversal" (HR.H
>> H.RH.), or does it mean that we should leave the PIE
> reconstruction at *HRH?
I don't think we can left PIE syllabification as undecided in such
cases. The well-known shared oddities of syllabification reflected in
the branches must be primitive rather than derived independently. The
behaviour of *CRHV in Greek is the same as elsewhere, cf. barús : Ved.
gurú- < *gWr.h2ú-. Greek diverges from other branches in those cases
where there is a potential conflict between the vocalisation of
(syllabic) resonants and (hitherto non-syllabic) laryngeals. E.g. in a
structure like *tr.h1t- or *k^m.h2t- the vocalisation of the resonant
(by prop-vowel insertion, e.g. *r. > *&r or *r&) was older than the loss
of the laryngeal, producing *t&'r&1t, *k&'m&2t (> *t&1'r&1t-,
*k&2'm&2t-) under accent and *tr&h1t-, *km&h2t- (> *tr&1ht-, *km&2ht-)
elsewhere (eventually giving téret-, kámat- vs. tre:t-, kma:t-). The
syllabification reversal is apparent. We have laryngeal "breaking" in
preconsonantal ih2/3, uh2/3, most likely via *ih2 > *i&h2 > *i&2h > ja:,
with a shift of syllabic prominence within the centring diphthong. The
assumption of a secondary character of the Greek developments is borne
out by Hinge's recent observations concerning the loss of preconsonantal
laryngeals in Greek (from the 2005 Cambridge conference). An initial
laryngeal does _not_ develop a Greek prothetic vowel if the next
syllable is closed in terms of _PIE_ syllabification. If something like
*h2wlh1nó- ends up as Gk. le:nos, the original syllable division must
have been *h2wl.h1$no-, with a syllabic lateral and a consonantal *h1.
> What does HRHV give in Greek?
I would predict *HR.HV- > *H&RHV- > VRV-, with the colour of the "schwa"
dependent on the initial laryngeal. This means, unfortunately, that
*HeRHV- yields the same outcome and the two are difficult to
distinguish, so the best examples should involve the zero grade of roots
of the form *HReH-. Perhaps amao: (if from prevocalic *h2m.h1- + -ao:)?
Piotr