On Mon, 22 Jul 2002, Miguel Carrasquer wrote:
> [On the causative]
> As for me, the reason for believing that *-ey{e/o} is in fact an
> incorporated verb is as follows:
>
> * In Sanskrit, -aya- alternates with -paya- as the iterative/causative
> suffix, -paya being used after verbal stems ending in a laryngeal
> (i.e. in a vowel, after the loss of laryngeals).
>
> * If -aya- and -paya- are variants of the same entity, and if the -p-
> is not a feature of the preceding verbal stem, then p- must be a
> prefix.
>
> * The only prefix I know in PIE of the shape p- is the preverb *pe(:)-
> / po-.
>
> * Therefore, the entity *ey-{e/o} must a verb.
>
> * The Hittite thematic verb iyami can be derived from *ey-{e/o}-mi,
> and the semantics ("do, make") are impeccable.
I've got to hand it to you, Miguel: Even if this analysis may one day turn
out to be impossible, it has a distinct touch of ingenuity. It is one of
those suggestions every reconstructive linguist would be proud to have
made.
For the suggestion to gain credibility (if it doesn't have enough
already), one would like to see the -p- in causatives outside Sanskrit (it
is not even seen in Iranian, but that may just as well be due to loss).
And one would like to have comparable prefix solutions for other
extensions of the causative, such as Lith. -dyti and Slavic -viti. If
there is found a set of preverbs that would work here, the theory would be
correct beyond a reasonable doubt. But the -v- is rather patently from the
root *bhuH-, and -d- may quite well stem from the reduplicated verbs *ded-
and *do:d-. Therefore, as long as the PIE age of the Skt. -p- is not
proved a fact, one would rather favour a solution by resegmentation - but
then, of what? I have checked va:- 'blow' and vap- 'scatter' which both
form va:páyati, but there seems to be absolutely no common ground in their
pragmatic semantics, even including compounded forms.
And it sure is evidence that the verbal root that lends itself the best
by standards of form and meaning is also found with a preverbs that fits
the variant with the the mysterious Sanskrit -p-.
> Of course that is only half the story on the causative/iterative
> formation. The other half involves the peculiar shape of the verbal
> stem (o-grade or zero grade), for which, as Jens will perhaps be happy
> to know, I now accept the solution advocated by him, i.e. an infix
> (originally, and before *r- still so, a prefix) *R, which caused the
> root vowel to shift to *o (in light roots) [in my view: lengthened the
> root vowel], and caused laryngeal deletion if the root contained a
> (final) laryngeal. The same element *R- plays a role in nominal
> (thematic) formations of the type *togah2, *bhóros and *bhorós.
Sure I'm grateful for this support. Yet, being the bastard I am, I am not
too pleased with what you add to the theory. Especially, I just won't
accept the statement that the prefix-turned-infix influenced the root
vowel. In the basic types, the root vowel is gone because the (originally
unaccented) root is in the zero-grade, and it is precisely the
accumulation of asyllabic elements that caused the reductions observed, as
-mn- to -m-/-n- and deletion of laryngeals. Thus the -o- IS the infix
consonant itself in syllabified form. A uvular or pharyngeal spirant being
vocalized as /o/ is not unheard-of.
> Until recently, I was reluctant to accept any of this, in part for
> aesthetic reasons (not only do we have to introduce a new phoneme *R
> but we also have to accept an "ugly" infix), but more importantly (as
> I now see) for lack of justification of what the _meaning_ of this *R-
> would have been.
>
> I now have an idea about this, which I offer for Jens' consideration.
> I realized that the formation of causatives sometimes (or often, for
> all I know) involves the root of the verb "to be". Two cases known to
> me are the Georgian-Zan causative prefix *ren-, *rin-, which contains
> *r- "to be" (Klimov), or the Basque causative prefix e-ra-, which may
> well contain the root *da "to be" (if from *e-da-). Without too much
> effort, we can also imagine the copula being present in nominal
> formations like **R-bher-os > *bhoros "which IS carried / which IS
> carrying". Even the o-grade of the perfect/stative (*woid-e "he IS
> (in a state of) knowing") may ultimately have the same origin, even
> though the behaviour of that *o is not quite like the behaviour of *o
> in the causatives and the aforementioned thematic formations.
The IE perfect does not contain the infixal -o-, for the perfect ablaut ó
: zero depending on accent, which is quite different from the caus.
*mon-éye-ti. The functional assessment is very difficult, for a number of
reasons: What is half a causative morpheme supposed to mean? Note that the
causative is formed with the infix vowel *and* a suffix. What would it
mean if it produces verbal nouns when combined with a suffixed thematic
vowel and is put in the collective, as in toga? It looks like it's forming
adjectives (vìlna : *wól-no-s > Gk. oûlos 'woolly'), in which case the
infix means very little, for the thematic vowel also has this force alone.
Since it originally precedes the word, could it perhaps be a word in its
own right, something like an article combining the adjective with its head
noun? Of course the semanteme of "being" would also fit, by why would
adjectives not always have that? The obligatory use in agent nouns
(*bhor-ó-s, no type **bhr-ó-s) would then be like the n-stem form of the
Germanic weak adjective. The basic difficulty is of course that the
lexicalized remains have mostly survived only because they took a semantic
turn and so were not obliterated when whatever they were originally
designed to mean was replaced by new and productive derivatives. That
means that most of the time the semantic shade is unoriginal.
Jens