What happened to Pre-IE's inanimate thematics?

This past week has been a little busy (i.e. a new job, looking after my parents' bratty dog, summer tanning, etc.) and so I've been slacking a little on my blog. Mea culpa. Of course, hectic life doesn't stop me from thinking and theorizing about ancient languages. NEVER! It only stops me from typing for a while. It's probably just as well because it gives me time to develop some more elaborate ideas to share. So I just noticed something in the past few days that I had never realized before concerning inanimate thematic nouns.

Over the years, I've slowly come to the conclusion that Syncope didn't obliterate all the schwas in early Late IE (eLIE) and that a few resisted the rule due to reasons regarding proper syllabicity. So a disyllabic wordform in Mid IE (MIE) of the shape *CVC.CV- should have preserved its final schwa because if it had disappeared, it might create problems with the distribution of the consonants in the remaining syllables^[1]. For example, I believe that the PIE word *h₁éḱwo- "horse" had survived Syncope (i.e. MIE *ʔékwa- > eLIE *ʔékwə- and not **ʔéku-). I also concluded thus far that some animate thematic nouns were derived from genitive-declined stems that are an artifact of a stage of Pre-IE when adjectives were once formed by way of the genitive case (either *-ás or *-ám at the time) and did not originally agree with the case of the nouns they modified. Inanimate adjectives were declined with *-ám while animate adjectives were declined with *-ás until they were "misanalysed" by prehistoric speakers as vowel-ending stems with terminating consonantal endings, nominative *-s (e.g. *wĺ̥kʷo-s "wolf") and a (nomino-)accusative ending *-m (e.g. *yugó-m "yoke"). Or at least, so say I.

That being said, I was recently investigating a potential loanword from Semitic, namely the "horn" word. In Proto-Semitic, the word is *qárnu while in Proto-Indo-European it's *ḱr̥nóm. At first, I was becoming disheartened by the apparent incongruence in accent, phonetics, etc. However, when I allowed my mind to explore what might have theoretically happened to thematic inanimate stems during Syncope, I came to realize that my theory had a gap that needed filling out. You see, if one ponders what would happen to a inanimate, rather than animate, vowel-ending stem in MIE during this period of time, we have to conclude that the final schwa should be preserved in the same way as animate stems when of the shape *CVC.CV-. If so, we end up with an interesting grammatical possibility because since inanimate nouns were endingless in the nominative or accusative case, my theory suggests that there should have originally been inanimate nouns ending only in a schwa.

So let's say that Proto-Semitic *qárnu was borrowed into Mid IE. We might rationally expect a form like *kérna, more than anything. Syncope would then fail to affect the final vowel in order to avoid syllabicity issues, thus a bare inanimate wordform *kérnə in the nominative and accusative cases. Obviously, at this point, this is not PIE as we know it but once I traced this interesting hypothesis to this point, I realized that I might be able to fill in the rest of the diachronic development. I think what happened next was that, for whatever reason, bare stems were undesirable to speakers and that once former genitive-declined adjectives were misanalysed as thematic stems, this provided a means to "complete" the awkward wordform with an imaginary "inanimate nomino-accusative" pseudo-ending *-m. Thus, this stem as well as other inanimate thematics must have been converted to this new "genitivized" form, thus *ḱr̥nóm with accent on the final syllable in imitation of genitive plurals. As extra food for thought, if this conversion had not taken place, inanimate thematics in Proto-Indo-European might have ended in *-e instead.

NOTES
^[1] These stem-final schwas are not always preserved and it appears to depend on individual phonotactic considerations. For example, I reason that MIE *ʔékwa- "horse" and *kérna "horn" would likely have resisted Syncope because *kw and *rn wouldn't have been valid syllable codae in early Late IE, the former violating sonority hierarchy and the latter showing adjacent tautosyllabic resonants which would have offended language-specific phonotactic restrictions at the time. On the other hand, MIE *nakʷta- "night" appears to have been reduced to early Late IE *nakʷt- (> PIE *nokʷt-) without interference, but then again, this is natural since the resulting cluster *kʷt was a perfectly valid syllable coda.

UPDATES
(Aug 09 2008) Added an important footnote about the phonotactic conditions behind stem-final vowel loss or preservation.