elmeras2000 wrote:
> Could you explain the nature of such a process?
I'd imagine it as follows:
The pre-PIE arrangement of non-high vowels was vertical without a
front/back contrast, i.e. there were two "true" vowels differing in height:
[-low] *&
[+low] *a
Their distribution was unbalanced: *a was restricted to root morphemes,
while *& could occur freely also in derivational affixes and
inflectional endings. This fact alone made *& by far the more frequent
vowel quality. A possible reason for such an asymmetry would have been
root stress at some stage in the development of pre-PIE, permitting the
full range of vocalic and consonantal contrasts in roots but
constraining it in other morphemes. As is often the case with
mid-central vowels, *& developed several allophones, and eventually
split into two mid-vowel phonemes, front *e and back *o (the latter
originally the conditioned member of the pair). Various phonological
processes produced also their lengthened counterparts (*e: and *o:) in
certain environments, and caused the elision of the vowel in others. On
the other hand, *a could be lengthened or lost, but did not alternate
with other vowels.
Patterns of alternation involving *e as the fundamental vocalism
alongside derived o-grades and nil-grades became grammaticalised, so
that the choice of a given grade depended largely on its morphological
function. Roots containing old *a remained outside the ablaut system,
and rare alternations like *na:s-/*nas- were perceived as aberrant
(partly because their defective distribution: they did not occur in
suffixes). Original a-roots had no o-grade derivatives, and there may
have been a tendency to get rid of their lengthened and reduced
allomorphs, generalising constant a-vocalism instead. A number of such
roots survived as a marginal component of the PIE lexicon, but otherwise
the PIE morphophonological system (and especially derivational
morphology) was pervaded and dominated by ablaut rules defining the
various productive transformations of fundamental *e.
Compare with the above the later tendency to eliminate ablaut from roots
with *e coloured (more recently) by an adjacent consonant. Alternations
such as *h2ak^- (< *h2ek^-)/*h2ok^-/*h2e:k^- are attested but
infrequent, while those like *kan- (< *ken-)/*kon- or *kap- (<
*kep-)/*ke:p- are extremely rare indeed. Other minor patterns (such as
Narten ablaut -- also, like "independent *a", restricted to root
syllables) were likewise marginalised.
To sum up, it's possible to view "independent *a" as a peripheral
archaism rather than an innovation. If this scenario makes any sense,
PIE came very close to being a monovocalic language at a sufficiently
abstract level of phonological description (its derivational morphemes
can be so analysed, and so can _most_ roots), but the messy residue of
an earlier stage survived at least in the phonemic makeup of root
morphemes, whose fundamental vowel may have been something else than *e.
Piotr