The material
Skt. ná:ma, G. ná:mnas
Arm. anun, G. anuan
Grk. ónoma, ónuma, énuma, G. onómatos
Alb. emër ~ emën
Lat. no:men, G. no:minis
Umbr. nome, Abl. nomne
OIr. ainm(m) n-, G. anm(a)e
Goth. namo:
OE nama (m.)
OFr no:mia "to name"
OPr emnes
Slav jImeN
ToA ñom
ToB ñem, pl. ñemna
Hitt. la:man
Discussion
As shown by Arm. (a-) and Grk. (o-, e-), the word started
with a laryngeal. Since Greek o- may be the result of Umlaut
(enoma > onoma), e- has more chances of being original, and
therefore the laryngeal was */h1/-
Next comes */n/. Hittite l- is likely to be the result of
dissimilation (*na:man > la:man). Tocharian ñ- at first
sight would appear to indicate a root vowel *e: (+h1ne:mn >
ñem/ñom), which is impossible, but if the initial laryngeal
is indeed *h1, Toch. ñ- may simply be a special development
of *h1n-, cf. *h3n- > m- in the word for nail *h3noghW- >
TochA. <maku>, B <mekwa>.
The root vowel varies between /o/ when stressed and zero
when unstressed.
Whether there was an additional laryngeal after the root
vowel is a matter of some debate. Sihler, for instance,
denies it, explaining the /o:/ in Latin <no:men>,
<cogno:men> by analogical influence from *g^noh3- "know".
Skt. /a:/ can be from *o (Brugmann), Armenian /u/ likewise
(*o > u before nasal), Hittite /a:/ can show normal
lengthening of a stressed vowel. The laryngeal is only
required to explain the Germanic verbal forms with /o:/
(e.g. OFr. <no:mia> above). At the same time short /o/ is
shown directly by Greek, Germanic, Umbrian and perhaps
Tocharian (PIE /o/, /e:/ > PToch. /e/, while /o:/ would
rather give /a:/).
However, it's hard to imagine that the suffix in this word
would _not_ be the ubiquitous *-men, and if so, that would
leave the root as an impossible *h1nV- (no PIE root can be
CCV). Also, Sanskrit <na:ma> _can_ come from *no:mn.,
Hittite /a:/ _can_ come from long /o:/, the development of
PToch /e/, /o/ and /a:/, especially in the neighbourhood of
nasals and labials, is not quite clear, the reduction of /m/
in Armenian <anun> is easier to understand if we depart from
*/ano:mn./ rather than */anomn./, Greek /o/ can come from
vocalized *h3, and so could Germanic /a/. Umbrian <nome>
(*/o:/ > /u:/ in Oscan and Umbrian), assuming it's from a
late inscription in the Roman alphabet (the old Umbrian
alphabet had no /o/) _could_ be due to Latin influence.
I would therefore prefer to reconstruct the root as
*h1nóh3-, oblique *h1nh3-, with suffix *-men-. The paradigm
would have been typically amphidynamic, and is faithfully
reflected (apart from the levelled root shape na:m-, and the
fixed accent) in Sanskrit:
NA *h1nóh3-mn. = ná:ma
G *h1nh3-mn-ós = ná:mnas
L *h1nh3-mén-i = ná:mani
pl *h1nóh3-mn-h2 = ná:ma:(ni)
Likewise Arm. anun < anú:(w)an < &no:mn., G. anuan <
anu:(w)ános < *&1no:mn.(n)ós [or, as Olsen suggests,
replacement of the suffix *-m(e)n- by *-m(e)nt-, as in
Greek, and special development of -nt-]; Latin no:men,
no:minis; Hitt. *no:mn. > *na:man > la:man, etc.
In my view, this reflects earlier:
NA *hnú:x-man
G *hnu:x-mán-âs
L *hnu:x-mán-a
pl *hnú:x-man-x
(or perhaps *tnú:k-man, etc.), with the regular reflex of
the pre-PIE vowels (**u: > *ó ~ zero, **a: > *o, **a > *é ~
zero), labialization of *k/*x to *xW = *h3 after *u(:),
amphidynamic stress-shift in the G (and Ins/Abl) sg.
(*hnu:x-mán-âs > *hn&x-m&n-á:s > *h1nh3-mn-ós), etc.
Three additional questions remain.
(1) The vocalization of laryngeals and nasals in the zero
grade forms appears to be erratic (perhaps we can expect
nothing else in an initial cluster *h1nh3-mn-): we have
Greek *&1n&3mn.(t) > <onoma> with vocalization of both
laryngeals, Germanic *n&mo:n > <namo:> with vocalization of
the second laryngeal, and OIr. and Slavic *n.(:)men > <ainm>
~ <jImeN> with vocalization of the nasal.
(2) Why do we find so many paradigms with the zero-grade
forms generalized throughout? If the above analyis is
correct, Greek, Albanian(?), Celtic, Germanic and
Balto-Slavic have replaced a perfectly sensible NAsg.
*h1nóh3-mn. with a NAsg. reflecting *h1nh3-. Why?
(3) The Old Irish genitive anm(a)e is thought to reflect a
PIE genitive form *n.(:)mens (*h1nh3-men-s). Final -e can
indeed not be explained in any other way. But a G.
*h1nh3-mén-s is unexpected, and certainly doesn't follow
from the amphidynamic paradigm postulated above. Such a
form could only have arisen in a neuter "collective" of the
type with long vowel in the suffix (cf. for instance *wódr.
~ coll. *udó:r). We would have had:
NA *hnux-má:n-x > *h1nh3-mó:n = Goth. namo:
G *hnux-ma:n-ás > *h1nh3-mén-s = OIr. anm(a)e
L *hnux-ma:n-á > *h1nh3-mén(-i) (as in the AD paradigm)
Now such a paradigm would not only explain OIr. anm(a)e, it
would also answer question (2). The zero-grade NA singular
root forms come from this "collective", even if (outside of
Germanic) the suffix/ending of the NA sg. itself has usually
reverted to *-mn., from the "normal" amphidynamic paradigm
(Grk. onoma, probably OIr. ainm --if from *n.mn., but could
also be *n.men--), or oblique *-men- has been generalized
(maybe OIr., OCS imeN, perhaps Alb. imën ~ imër --or is that
from *-mn.?--).
=======================
Miguel Carrasquer Vidal
miguelc@...