IE thematic vowel rule

The Thematic Vowel Rule: Saussure had it already.

The quarrel over the e/o alternation in the Indo-European thematic
vowel has gone far enough. I have been called all sorts of
flattering names, and so have the facts. I do have an embarrassing
confession to make, viz. that I have now seen that the voice-
governed thematic vowel rule is not at all my own discovery, for
Saussure had it already.

Before I go on, may I say that Saussure's Mémoire makes for very
hard and tiresome reading, being filled with a lot of "noise" in the
particulars which just make the modern-day reader's mind wander.

Still, I am all the more embarrassed to see this, for I *have*
indeed read (many parts of) Saussure's Mémoire, but I had apparently
formed myself a wrong impression about what the message was in this
particular case. I remember checking on the Mémoire some time in the
nineties and getting the impression that Saussure advocates
*rounding* in the following segment as the cause of the
transformation a1 > a2 (what we call e > o), because he derives 3pl
nt from earlier mt. Now, on renewed inspection, that is exactly what
he does NOT do.

In his presentation, Saussure addresses the thematic verb, but later
brings in nouns and pronouns too. I give a few quotes (pages of the
original of Leipzig 1879, in the Genève 1922 reprint the page
numbers may be a few pages lower):

p.86:
Les langues européennes montrent clairement que la voyelle ajoutée à
la racine dans les thèmes verbaux en -a est un a1 qui alterne avec
a2. Il y a concordance de tous les principaux idiomes de la famille
à la place où apparaît a2 (1e pers. des trois nombres, 3e pers. pl.).
...

p.87:
La forme primitive exacte de la 1e presonne du singulier de l'actif
est une énigme que nous n'essayons point de résoudre. Avec la
désinence dite secondaire, elle n'offre pas de difficulté: gr. é-
pheron, sl. vezU (régulier pour *vezon), skr. á-bharam (a bref vu la
syllabe fermée). Du reste le paradigme se répète partout où il y a
une conjugaison de l'espèce qu'on appelle thématique. Dans ce
paradigme, l'apparition de a2 est évidemment liée d'une manière ou
d'une autre avec la nature de la consonne qui suit. .. On ne peut,
vu la 3e pers. du pluriel, - à moins d'admettre que la désinence de
cette personne fût à l'origine -mti - chercher dans le son labial la
cause de la transformation. Il faudra l'attribuer aux *sonantes*, ou
plus généralement peut-être aux *sonores*. C'est le seul cas où la
substitution du phonème a2 au phonème a1 trouve son explication dans
une action mécanique des sons avoisinants.

Saussure goes on to point out that there is also a2 (i.e. /o/)
before the suffixes of the optative, and of the participles in "-
mana-" and -nt-. On p. 90 he comes to the nominal o-stems. He notes
*-os, *-om, acc.pl. *-ons, and vocative *-e. Strangely, there is no
comment on the obvious discrepancy from his main rule offered by *-
os. He says this, however:

p.90:
Tout le reste est plus ou moins entouré d'ombre. Doit-on, au
*génitif singulier*, admettre a1 ou a2? Le goth. vulfi-s parle pour
la première alternative, le gr. híppo-io pour la seconde. [He then
mentions the pronominal forms Sl. c^eso, av. cahya and goes on:]
Comme il n'y a pas d'ailleurs de raison de croire que le génitif
d'un pronom en -a2 différât en rien de la forme correspondante des
thèmes [p. 91] *nominaux* en a2, nous concluons à l'indo-eur. akwa1-
sya et nous tenons l'o de híppo-io pour emprunté à d'autres cas.

His further examples include nom.pl. "akwa2 + a1s. Prononcée avec
hiatus, la forme serait akwa2a1s (à peu près ekwoes); avec
contraction akwa:2s (ekwo:s)." [91]. The ntr. of pronouns *tod and
the nom.pl *toi are correctly assigned a2 (i.e. *o) [92/93]. So is
the isolated nom.sg *so, but there is no comment on the strange word-
final *-o. He mentions, but does not explain, that first members of
compounds end in *-o-, but he does see the potential regularity in
the derivative type in *-eta:- of adjective abstracts.

There is no following summarising of the findings, so the reader is
left to himself to collect and sift the details. Even so, the
opening words that this is the only case where the language shows a
mechanical selection of -e- or -o- in obvious dependency of the
phonetic nature of the follwoing segment are so clear that I find it
somewhat chilling to realize that I have overlooked it for so long.
Other than others, however, I have not been fooled, for I have
derived the very same regularities from the facts, only on a broader
basis and with more details. But the exactitude with which this
simply anticipates what I have later seen myself is breathtaking.

I beg the moderators' indulgence when I choose to post this also on
Cybalist which is where the thread now belongs. It has lost all
connection with Afroasiatic and has become a purely internal Indo-
European matter. I shall respond to any reactions on Cybalist.


Jens