Intensive Reduplication

[Jens Elmegaard Rasmussen]

On Wed, 9 Jun 1999, petegray wrote:

> PIE *Hiugom and Greek initial z

> There is of course a continuing debate about the origin of the apparently
> double reflex in Greek of PIE initial *y-.    It is an attractive
> proposition to suggest *y- > Gk /h/, while *Hy > Gk /z/, but the evidence is
> rather more complicated than Nath suggested, and the idea is still debated.
> There are some who think the complications are too much, and the idea can't
> work.
[...]

Pardon my intruding, but I'd like to say that this still sounds
attractive, if not in such a clean fashion. The matter was considered an
open-and-shut case for a period of time in the mid-seventies after Martin
Peters presented the analysis of Attic Greek /hi:'e:mi/ from
*H1yi-H1yeH1-mi where the presence of a laryngeal could be proved. That
led Peters and the rest of us to conclude that *Hy- gave Greek /h-/ and
that, by contrast, Gk. /z-/ would have to be from plain *y-. Then came
Rix' Historical Greek grammar containing beautiful arguments for initial
laryngeals in some roots with Gk. /z-/ corresponding to Vedic y- and
lengthened augment or compositional vowel. That brought the matter back
where it started, except that Peters' /h-/ for *H1y- still stands.

Plain IE *y- may be seen in Ved. yaj- 'to sacrifice', of which Gk. ha'gios
'holy' is a gerundive formation, cf. lack of lengthening in the
desiderative i'yaks.ati. That indicates that, e.g., the relative pronoun,
Gk. /ho's/, Ved. ya's, is IE plain *yo's, and that IE *y- without
laryngeal gives Gk. /h-/.

Then the difference between Gk. /h-/ and /z-/ is not just one of *Hy- and
plain *y-, but we have to differentiate the laryngeals - after all we
accept three of them. As /z-/ is also the outcome of *dy- and *gy- (with
all kinds of g's), there would be very good sense in taking /z-/ to
represent also *H3y-, given the phonetic character of *H3 as a voiced
labiovelar spirant (or labiouvular or labiopharyngeal as some prefer on
grounds I do not see). There remains only *H2y- for which /z-/ and /h-/
appear about equally likely. Phonetically *H2 was something like [x], a
voiceless velar or postvelar spirant. If the coalescence of *y- and *H1y-
means that *y- was devoiced to a voiceless y (much like the initial
of English _huge_), then *H2y- [xy-]   could well be expected to yield the
same, i.e. /h-/. But if it was the stronger body of *H3y- that set it off
from *H1y-, one could just as well expect *H2y- to join the "strong"
series and end up as /z-/. The latter possibility, however, would mean
that the y part of *H2y exerted an assimilatory influence on *H2 by
voicing it, which is perhaps not very likely if *y- alone loses it voice.

This brings us to the picture:

   IE   *y- > Gk. /h-/
   IE *H1y- > Gk. /h-/
   IE *H2y- > Gk. /h-/ (or /z-/?)
   IE *H3y- > Gk. /z-/

Jens