Davush wrote: ↑
Sat 25 Nov 2017, 23:16
I am trying to think of some interesting ways word order can interact with case/definiteness. If a language is predominantly SOV, would the following be likely:
woman.NOM man.ACC saw
'The woman saw a man'
woman.NOM saw man.ACC
'The woman saw the man'
With a definite object appearing after the verb. Or would it be more likely to be the other way around in a SOV language? Is anything like this found in any nat langs? I know Mandarin sometimes has VS to convey indefiniteness.
Your example is actually similar to some things that do happen in some natlangs. Some of the phenomena I talk about below are specifically geared toward languages with dominant VO word-order rather than OV word-order.
(My memory being what it is, the remainder of this post is all 
What happens to an object depends on its animacy and its definiteness.
I once suggested a system, in a collaboration conlang, wherein:
Definite human objects had to come after the verb, had to be in the accusative case (if there was one) and/or have the accusative adposition (if there was one) and/or have the definite article (if there was one);
Definite animate objects, and specific/referential human objects, had to come after the verb and be in the accusative case and/or have the accusative adposition, but might do without an article (if it wasn't both definite and human);
Definite objects, human objects, and specific animate objects, had to come after the verb, but might not have case-marking nor any adposition (if it was definite but inanimate, or human but nonspecific/nonreferential, or specific animate but indefinite and nonhuman);
Specific nonhuman objects, and definite inanimate objects, could not only do without case-marking, adpositions, and articles, but could come before the verb; however they couldn't be incorporated in the verb;
Nonspecific/nonreferential inanimate objects could be incorporated in the verb.
The thing is, no particular piece of that system is unrealistic or unnaturalistic. The whole collection might be, taken altogether. But it's true that in languages where not all objects are treated the same, the more definite and/or the more animate it is, the more likely it is to not be incorporated in the verb and to be specifically marked for nominal features, in particular for its case-role, and for definiteness or specificity.
And in languages that are mostly VO, the less definite and/or less animate an object is, the likelier it is to move before the verb; and in languages that incorporate certain indefinite objects into verbs, they usually won't incorporate an object that couldn't appear before the verb if not incorporated, and won't incorporate it unless it is shorn of all inflection.