Personal Pronouns

[Jens Elmegaard Rasmussen]

On Fri, 23 Apr 1999, I found:


> [ Moderator's comment:
>   There is no evidence for a laryngeal in this stem; the length is due to
>   Ablaut ("apophony"), unexplained as yet.
>   --rma ]

It seems to be part of the _definition_ of ablaut that it is unexplained.
Well, I am at least trying to explain the ablaut. The whole matter is
simply one of internal reconstruction. There is no more ablaut in IE than
there is umlaut and breaking in Old Norse or Old English. Though that _is_
quite a lot, it has not kept scholars from successfully pointing out the
rules of co-variation which can often be detected by comparing different
parts of the language itself. In PIE there is a lot of such co-variation,
in that the vocalism manifestly varies with the accent and - to a more
limited, but specifiable, extent - with reduplication and special
morphemes. A stem-final vowel ("thematic vowel") is not lost, but varies
e/o according to rules of its own (see on next comment). If we find that
the nom.sg. in *-s is restricted to certain stem classes, while others
have zero, but they all have lengthening when the stem ends in a
consonant, an obvious _possibility_ is that this reflects pure phonetic
chnge. And if, on top of this, we find that some neuter stems ending in
consonant + s behave just as if their s were the nominative marker, which
of course it cannot be in a neuter noun, our guess is confirmed: Then
there is little room for any understanding except by simple phonetic
change. If that demystifies ablaut, so be it.

I also found:

> [ Moderator's comment:
>   It appears to be current in some European schools, that *e/o is [e] before
>   voiceless obstruents and [o] before voiced.  I find that there are too many
>   exceptions to accept this.
>   --rma ]

Oh yeah? That's nice, tell me where - for I have been preaching this
sermon since the mid seventies, encountering nothing but scorn and
disbelief, except with a few unbiased colleagues who bothered to get
informed and, it seems, now some very young Germans who want to look at
the facts for themselves with an open mind and are now finding the same
thing, plus own students who have heard enough about the matter. I have
later seen that Saussure actually formulated what I believe is the correct
rule, but only to give it up for a wrong and narrower formulation; he
believed the active feature to be rounding, deriving the 3pl *-nt from
earlier *-mt which is not possible (West Germanic specifically demands
*-nth, not *-mft which would have been the outcome of *-mth; and if the
rule is also to explain the active participle in *-ont-, Lithuanian
/-ant-/ proves n, not m) and of course contrary to forms like *-od and
*-oy which do not have rounding, but do have voice. - There ARE many
exceptions, but they are principled and harmless: This particular rule
applies only to the "thematic vowel", i.e. pure vowels in stem-final
position. It explains the interchange e/o in the stem final of thematic
verbs, where there are no sure exceptions at all, and in pronouns where
inflections retaining maximum alternation show the very same distribution
as the verbs. It explains only a few, but precious, forms of the nouns
where the alternation hardly occurs, presumably because of a simple
standardization of the variant *-o-. Even in the noun, however, we do have
*-e in the vocative (before zero, just as the verb has *-e in the
imperative), and the fem.-ntr.pl. in PIE *-aH2 must be from *-e-H2 since
*-o- can be proved not to get coloured by /H2/ (the verb plays along in
the 1sg.mid. *bher-e-H2-i > *bheraH2i, the necessary common point of
departure of Skt. bhare and Gk. pheromai). - Note that the rule also works
for a stem-final vowel ("thematic vowel") before another suffix, i.e. not
only before desinences: ptc. *-o-nt-, *-o-mH1no-, opt. *-o-yH1-, stative
*-e-H1-, factitive *-e-H2- (> *-aH2-), adjectival abstract *-e-taH2-
(Goth. diupitha which cannot be analogical; Gk. neo'te:s can). The rule no
longer applies to new coinings, thus the compositional vowel is a
standardized *-o-, just as there is *-o- in the hypochoristic suffixal
array *-o-ko-. It is like /s/ in Greek: Anybody can see that /s/ is lost
intervocalically in old forms (kre'as, gen. kre'a-os), but often restored
in younger forms (lu:'so:, e'lu:sa on the pattern of kle'pso:, e'klepsa).
Just as the Greek "s rule" may appear to be disproved by reference to
cases of actually occurring intervocalic /s/, thus the IE "thematic vowel
rule" could be seemingly disproved by its exceptions. Such counterevidence
only disproves the productivity of the rule, not its earlier existence and
its explanatory adequacy for a great many forms lying around in the
morphology and the lexicon.
- But thanks for the comments; this is the way to make progress.

Jens