On Sun, Jun 11, 2000 at 08:53:51PM -0500, Thomas R. Wier wrote:
> Nik Taylor wrote:
> But that's the thing:  a language that has only recently gained a case
> system would have grammaticalized (cannibalized, if you will) its
> adpositions to create that very system.  And, for the reason Ed has
> already mentioned, adpositions in languages with morphological case
> distinctions are usually used to  *specify* a further nuance of meaning
> already generally conveyed by the case itself.  They begin as adverbs
> or particles, being only optional, and slowly become obligatory through
> attrition of the case's use.  After a while, adpositions may help motivate
> the collapse of the case system to which they were originally only
> peripherally connected, and assume all of the cases' uses.  That is,
> at any rate, how it happened in many IE languages.

I'm not sure this is directly relevant, but look at how English has
overloaded some of its prepositions, either by piling them up or by
creating new "phrasal prepositions" out of preposition + noun +
another preposition -- "in front of", "in back of," "out of," "on to" ->
"onto", "in to" -> "into," "over to," "in the midst of" and so on.

If "to" or "of" became bound dative and genitive case markers
respectively, we would have a whole crop of new prepositions taking
the genitive, such as "infront" "inback," "out," "inthemidst" etc, and
others taking the dative such as the new "on+dative" which used to be
"onto," the new "in+dative" which used to be "into", "over," and so

I imagine that is at least part of how you go from an isolating
prepositional system to a preposition + cases system -- the most
commonly used prepositions, and especially the ones that can be
compounded in some way, turn into case endings, and the things that
used to be parts of "phrasal prepositions" (if you will) split off
into new prepositions.

At least, that's one way it COULD happen.

I imagine that after that you'd of course have a lot of analogical
shuffling, and prepositions that once did not have "of" or "to" would
be assigned to these new "cases" on the grounds of newly possible
semantic generalizations.....