Uralic and IE

[Jens Elmegaard Rasmussen]

On Wed, 7 Apr 1999, Glen Gordon wrote:
[Excerpts from discussion with Miguel Carrasquer Vidal:]
[snip]
> ME (GLEN):
>   First, whether the heteroclitic stems end in *-t or *-d changes
>   nothing since I've been saying that there was no pronunciation
>   contrasts in IE between *-t and *-d (or *-dh).
> MIGUEL:
>   Which is obviously false.
> ME (GLEN):
>   Obviously how?
> MIGUEL:
>   Sanskrit, for instance, has -d for the ablative, -t for the
>   3rd.p. sg.   Reason enough.
[snip: In reply, Glen invokes analogy with primary *-ti]

It should perhaps be pointed out that the choice of "*-d" as the
final consonant in the registration of the thematic ablative ending as
*-o:d (or *-oad to allow for Balto-Slavic contraction into /-a:-/) is
arbitrary: In Sanskrit sandhi, there is full neutralization of word-final
/t/ and /d/ (and /dh/ and /th/) in all combinations; Avestan has the
special (unexploded?) dental (transcribed by t with subscript tilde) for
both, and Old Persian drops both. Thus, Indo-Iranian offers no evidence as
to which dental consonant is underlyingly involved here. Nor does Italic
where *-t yields /-d/ in Latin and Oscan alike (and is lost in Umbrian).
In sum, there is no direct evidence at all.

There may still be indirect evidence, and that is to the contrary - much
as it would suit me to have a contrast to point to when I explain the
choice of /e/ in 3sg *bher-e-t vs. /o/ in neuter pronouns like *to-d
by the rule "/e/ before voiceless, /o/ before voiced" - :

If the ablative *-o-aD is a recomposition of stem-vowel /o/ + postposition
ultimately identcal with Slav. ot(U) 'from' (itself identcal with Lat. ad
'to', OE aet 'at', the semantic shading being caused by the case of the
following noun), then one may ask how the original form had been before
the recomposition took over. If the postposition had the shape *aD, it
would be expected to lose its vowel after the thematic vowel where all
ablauting morphemes show zero-grade. Now, depending on the voicing or lack
of it in the dental concerned, the result would be either *-o-d or *-e-t.
It now so happens that we do have a word *e-t, also extended to *e-t-i
which could very well be the ablative of the pronominal stem e/o-
(enclitic variant i-). The extended form is Lat. et 'and', Gk. e'ti Skt.
a'ti 'in addition, yet', while the unextended form is seen in Av. at (with
"funny t"), a frequent sentence opener of probable meaning 'and then'.
These all mean practically the same as the Skt. adverb tata's 'from there,
thereafter, thereupon, then'. If this is correct (I know it does not have
to be) the dental occurring in the ablative ending is a /t/ in the
morphophonemics of the protolanguage.

There is a case in Germanic of two word-final dentals being treated
differently in case of extension, retained to this day in English _that_
vs. _stood_. _That_ is Goth. _thata_ indicating that the dental was still
IE *-d or post-soundshift pre-Gmc. *-t at the time of the addition of the
particle sitting on the Gothic form. _Stood_ is Gmc. *sto:th (thorn
retained in Goth.) from *sta:t-e, a relatively simple reshaping of the
inherited root aorist *staH2-t made by adding the productive endings of
the IE perfect to the full form of the 3sg, thereby drawing the
desinential *-t into the synchronic root. In this case the once-final
dental is treated as IE *-t (or post-soundshift pre-Gmc. thorn). If both
are to be derived from phonological systems neutralizing
mode-of-articulation oppositions in word-final position, the neutralizing
habits will have to have changed between the two events. Such change is
entirely possibly - it could even be consistent: The result could have
been [-t] before the soundshift in *staH2-t > *staH2t-e and [-t] again
after the soundshift in *{kw}oD (*{kw}ot ?) > *hwat -> *hwat-o:. But it
could also simply be what it looks like, IE *-d and *-t being kept apart
all along. This is suggestive, but not entirely probative.

Jens