The sectors of ablaut.

To get over the stalemate in the quarrel over ablaut I shall have to
point out a number of things I do not find occasion to say if I only
respond to the posts of others.

It appears to be fundamental that the three main parts of the IE
inflected word have their own limits of vocalism:

Roots may have any vocalism, long or short; while by far the most common
root vowel is /e/ or /e:/, also /a/, /a:/, /o/ and /o:/ appear to be
represented. (I am not so sure about the long /u:/ of 'mouse' anymore.)

Suffixes only have /e/ (and what comes from that source).

Endings may have any vocalism, but only short: /e, o, a, i, u/.

Suffixes and endings may also be without underlying vowels.

The monotonous vocalism of the suffixes must be due to weakening, much
as German only has schwa outside the root syllable in indigenous words.
The invariably short, but still differentiated vocalism of the desinences
must have arisen in independent words which were reduced, if less so
than the suffixes.

Since some endings are vowelless, there must have been reductions going
on before the time of any ablaut we can uncover on the basis of IE
alone. At that time the suffixed stems must have been initially
stressed, since the suffixes have neutralized their vowels completely.
This means that the border between stem and flexive was once a
word-boundary separating a main word from a following "grammatical
word".

All of this has no necessary bearing on ablaut in the normal sense of the
word. In general terms, IE ablaut is the result of weakenings and
lengthenings that postdate the creation of the tripartite structure of the
inflected word. In *pér-tu-s, *pr-téw-s 'ford' there is full concord
between the accent placing and the ablaut, but in *p&2-tér-m/*p&2-tr-ós
there is underlying accent on the suffix with the neutralized vocalism.
Forms like *p&2-tr-bhís, *p&2-tr-sú (irrespective of the question whether
they changed to second-syllable-accented *p&2-tr'-bhis, *p&2-tr'-su
already in PIE) confirm the varied vocalism of the endings which is not
shared with the stem-forming suffixes. Rules of a nature we do not know
must at some remote period have changed something like *péh2y-tVr- into
*peh2y-tér- while leaving *pér-tew- unchanged (or changing something else
into that by even more enigmatic rules). Only after the consitution of the
stems *peh2y-tér- and *pér-tew- did inflectional accent go to work and
shift the accent to the following syllable every time the added flexive
synchronically formed a syllable. Thus there was no accent shift in the
strong cases in *-z, -Ø, *-m, MF.du. *-h3, nom.pl. *-zs, acc.pl. *-ms,
ntr.du. *-yh1, or coll. *-h2, nor in the active sg. verbal endings *-m,
*-s, *-t, while there was accent movement before all the other endings
which formed syllables. Half-way on its course toward zero, unaccented /e/
was lengthened before *-C-z and so became *-o:C-s in the end; the
corresponding accented segments of course became *-é:C-s, but that is less
telling. Before the nom. *-z fell in with the phoneme /s/, it was lost in
the endings *-V:nz, *-V:rz, *-V:yz.

The origin of the "thematic vowel rule" must be projected back to a
time when there was a palpalble boundary between stem and flexive. That
makes it older than the bulk of the ablaut rules, so the process is
only marginally related to the whole ablaut business. The thematic
vowel is simply a stem-final vowel, i.e. a vowel final in the segmental
sequence consisting of root + suffix(es). If the segment following the
boundary to the flexive began with something voiced, the thematic vowel
became /o/, otherwise it became /e/; it was not affected by the accent
at this time, being e/o no matter whether it was accented or not, and
notably not being lost when unaccented. A fine parallel would be Latin
which has unreduced first syllables, highly reduced interior syllables,
and retains most final vowels.

Throughout the period of the ablaut changes in the traditional sense of
the term the thematic vowel stayed unchanged. The ostensible initial
accent on the old stems makes it unadvisable to regard the thematic
vowel as originally accented. The old pronunciation of the once-final
vowel must have been retained as some feature in the vowel, be it a
tone, a secondary accent or semi-length, or for that matter a glottal
stop as my guess was at the very beginning. I have no strong feelings
in this, only I don't like to speak of length where I see no length
*oppositions*.

A stem ending in the thematic vowel could be further extended by
suffixes and so did not always occupy final position of the full stem.
The thematic vowel rule also worked in the position before added
suffixes, as opt. *bhér-o-yh1-t, but stative *séne-h1- 'be old'. Thus,
these stem-final or suffix-final vowels must have retained their
special feature down to a time following the main ablaut reduction by
which /-eh1-/ was reduced to /-h1-/ because not accented and then
selected /-e-/ as the form of the preceding thematic vowel.

Thus the *working* of the thematic vowel rule belongs to the time
perspective we can handle, but the *causation* of its special status
must go back beyond that time. But then again, so do the causation of
the main word-structure rules. This means, of course, that the
connection with outside families does not lie right around the corner,
which implies, in turn, that priority should not be given to external
evidence where internal evidence is available.

The specifics of my ablaut rules only differ marginally from those
propagated by Miguel, and they also agree in many respects with points
made by Glen. Despite the admixture of details I cannot accept I will
certainly not be ungrateful for the effort invested by them in
propagating the basic insight they contain.


Jens