ps

[Tom Givon]

PS:

Speaking of extreme typological cases, how about Southern Arawak 
(Machiguenga, Asheninka), where ALL SR's ('thematic relations') are 
coded on the verb even in MAIN clauses (except for one skinny, bleached 
locative preposition)? So you can ONLY say the equivalent 'I talked-to 
the-woman', 'I worked-with the-knife', 'I ran-from  the-house', etc. 
Never on the noun, only on the verb. Does it fit the zero-anaphora 
pattern? Well, if you think for a minute, it does too. S.A. does have 
pronouns, but 'discourse' zero-anaphora is most prevalent. So when you 
say high-frequency things like 'I talked to her' or 'I worked with it', 
schematically you have, in context, ['I talked-to  0'] or ['I 
worked-with 0']. In other words, in these high-frequency anaphoric 
expressions there is no place to hang the adposition BUT on the verb. 
And the diachronic process of  cliticization is driven by these 
frequencies (and adjacencies, and word-order).

The upshot of this is that we can classify atomic facts and quit there. 
Or we can try to classify them within broader patterns( of facts!) that 
show wider, interesting commonalities, and then look for some 
explanatory principles. The second mode of classification is admittedly 
more ambitious, so if it doesn't turn you on, sorry. (There is an 
offensive expression in Hebrew I won't cite here, straight out of 
Ecclesiastes. Eitan Grossman would identify it, I'm sure). But maybe 
ambitious stuff is too much, maybe it's not your stuff. See, this second 
mode of 'classification' is highly theory-dependent, it is not as 
theory-neutral as some people might prefer. It strives, in science in 
general, not only to describe but --at the same time--explain, through 
the constant interplay between data & theory. Yes, we've had this 
discussion many times  before. We seem to be hung up on an impoverished  
brand philosophy of science. Cheers,  TG