A form *dyeh1-yu would be analogous to *h2we1-yu, cf. Sanskrit Va:yu, the Wind God.
DGK: No laryngeal is necessary
if we follow the explanation of the long vowel given by Sihler (New Comp. Gr.
of Grk. & Lat. §§324-7). In his
treatment, the PIE acc. *djéum regularly became *djé:m (whence Skt. _dyá:m_,
Old Lat. *die:m, Old Grk. _Zên_, in Homer and Hesiod only verse-final with a
subsequent vowel, as though it were Class. Grk. _Zêna_ elided). This is parallel to the acc. *gWóum 'head of
cattle' becoming *gWó:m, with both acc. forms acting back on the nom. *djéus,
*gWóus to yield *djé:us, *gWó:us. The
long diphthongs are preserved in the Skt. nom. *dyáus, *gáus but underwent
Osthoff's shortening outside Indo-Iranian (with the possible exception of
_Ze:us_ in an inscription of Thera, but this is more likely a carver's error,
eta from the acc. inadvertently replacing epsilon).
An alternative to the
usual view that *djé:us comes from a /w/-extension of PIE *dei- (Pokorny, IEW
184) or simply from *deiw- 'to shine' (Watkins, AHD App.) is that it is an
active root-noun compounded from *dei- (regularly in the zero-grade) and the
simplex *jeu- 'to join, connect' (cf. *jeu-, *jeu@-, *jeug- 'verbinden', IEW
508, and Sihler's comments on nasal-present infixation, NCG §453). The protoform would be *di-jeu-
'light-joiner, one who connects the world with light, illuminator of the
world'.
In his criticism of Edgerton's
and Lindeman's Laws (NCG §§179-81), Sihler observed that the Rig-Vedic scansion
_diyáus_ occurs far less frequently than predicted by Edgerton, and exhibits a
distributional preference which Lindeman cannot explain. All but one of the 26 examples of _diyáus_ in
the RV occur in line-initial position. If
the protoform was in fact *di-jeu-, these disyllabic nominatives are archaisms,
already superseded by _dyáus_ at an early stage of Vedic, and their
distribution has nothing to do with Sievers' Law or any elaboration thereof. Greek likewise has only a monosyllabic
nominative. But in Italo-Celtic it
appears that the disyllabic acc. *dije:m generated a new paradigm for 'day',
with nom. *dije:s reflected as Old Irish _dïe_, Lat. _die:s_.
The Phrygian acc. _Tian_ also looks like a disyllable reflecting *dije:m.
I would thus suggest a
modification of Sihler's scenario, in which the original acc. *dijéum became
*dijé:m and acted on the orig. nom. *dijéus to create *dijé:us. Absorption of the */i/ from nom. and acc. did
not occur in PIE, but independently in Sanskrit and Greek.
Finally, _Deipáturos_,
according to Hesychius a god of the Tymphaeans or Stymphaeans (by either name
Illyrians), should not be identified with Zeus as is done by Pokorny (IEW 184)
and Krahe (Sprache der Illyrier 54). The
first element cannot plausibly represent a case-form of *D(i)jé:us. More likely it continues the vocative of a
root-noun simply derived from *dei- and the name is to be understood as 'Father
Light', not 'Father Sky'. The latter
would have been glossed specifically as _Zeús_, not generically as _theós_.