prepositions on finite verbs

[Tom Givon]

I need to apologize to Paul Hopper & y'all about replying in a hurry
before reading Claire's note on the Australian situation, which reminded
me of something I should have taken into account earlier--that there are
at least two well-known mechanisms for 'adsorbing' adpositions onto
verbs that have nothing to do with non-finiteness or nominality. One of
them is the grammaticalization of adpositions as "promotional" affixes
on verbs (so-called "applicatives"). I haven't read Claire's
dissertation, so I don't know if the mechanism she found is the same,
but in many Bantu languages this starts in REL-clauses (and passives, at
least in Bemba...), where the locative prepositions pa-/ku-/mu- become
obligatory V-suffixes in constructions quite analoguous to the English:
        'the woman I work for', 'the man I talled to', 'the house I live
in' etc.

In some Bantu languages (eg. KinyaRwanda) these become more general
(optional) promotional suffix when the LOC-object is topicalized &
becomes the DO of the clause (see Kimenyi's dissertation, UCLA, 1976).
In Rwanda, this is optional in main clauses but obligatory for the
relativization and passivization of obliques (non-patients). This
suggests that the process started in REL and PASS clauses (more general
in  Bantu; Meeeusen reconstructed a V-suffix in Bantu relativization, I
think), but later extended to main clauses in only KinyaRwanda and a few
others. Of course, not all "applicative" affixes come from adpositions.
Some come directly from (serial) verbs. The KinyaRwanda situation is
rather hetrerogenous, but at least one other promotional suffix -- the
-na used to promote associative objects, is also the associative
preposition.

Another mechanism that may or may not be partially related to the above
is the profusion lexicalized PREP-V verbs in both Germanic & Romance.
The English V-PREP lexicalized verbs ('break up', 'break-out',
'break-off' etc.) is probably a later example of the same or similar
mechanism. The closest case I know where this has been described
synchronically in a language where the process is in the midst of
happening is in Colette Craig/Grinevald work on Rama. The mechanism
there is not 100% clear, but an unpublished  text-based paper on the
Rama data (by Bonnie Tibbits) suggests some kind of zeroing of an
erstwhile PP (postpositional phrase), with the adposition surviving and
adsorbing on the adjacent verb. Bonnie's data suggested that this was
not anaphoric zero (highly referential), but rather an "antipassive"
zero (non-referential). One way or another, I should have remembered
these.

Best,  TG

=================