md0 wrote:I think that the third definition of Case is case as described by Indo-European philologists after all. Which doesn't mean much when you are discussing syntax.
Which makes a lot of sense if you don't consider it a syntactical category in the first place. I mean, gender doesn't mean much either when you are discussing syntax, and it's not a problem, because it's not meant to describe syntax.
I probably just fail yet to see the reason why the definition of a case based on syntax is useful, but the only reason I could think of to do this is if you wanted to postulate for some reason that case is univseral.
md0 wrote:So yeah, the answer is, as far as I can tell, that yes, under this model Nominative and Accusative are present in all languages that have thematic roles (which as far as we know, that's all human languages).
Okay, with that explanation, I see why you would be surprised if a language had a genitive, but not an accusative.
Only nominative, accusative and ergative. My understating is that this is an important distinction for the theoretical syntacticians: syntactic case (Nom, Acc, Erg) and idiosyncratic case (eg genitive, dative, locative etc) behave differently in various circumstances.
That feels rather inconsistent, though. Why don't you keep case and syntactic role separate instead of postulating that case is universal and making only some of them special just so you can use (abuse?) them for discussing syntax? Doesn't that overload one term with multiple separate meanings?
dEhiN wrote:My understanding and everything I've read up to now basically uses case in only a morphological sense for languages that have inflection and use those inflected (case) forms to denote syntactic function more so than sentence word order.
I don't think you can go far if you try to map 1:1 a morphological form and a syntactic function. In the simplest scenario it's one-to-many, but it's often many-to-many.
I wouldn't understand "denote" as a 1:1 mapping. But maybe it's more precise to say that the syntactic function influences the morphological forms, and therefore you can often infer the syntactical function from the case of the form.
I would be interesting to say if there are morphological accusatives that are not assigned by verbs to their objects. Because if I understand the arguments of the theoreticians well, that's the thing: syntactic case is only assigned by a verb (accusative to its complement, nominative to its specifier, in a simple configuration). Idiosyncratic case can be assigned by other words (a certain proposition might assign dative, nouns can assign genitive and so on).
But prepositions can certainly assign the accusative as well?
And the first thing I actually thought of for some reason was the Accusativus Graecus in Ancient Greek, which means something like "regarding ..." iirc.
tl;dr: linguistics and philology share many words, but have different definitions for them. I think for Ipse's question, the syntactic definition makes more sense, because he was wondering if there's a dependency between cases and there's already a parametric approach to syntax. A more philological definition of case that relies on form cannot be used to see if there's a dependency relation, because the same form of a noun might be used in many different environments.
I didn't understand Ipse's question to be about this syntactic thing, but that's something for him to clarify.
But that the same form of a noun might be used in many different environments doesn't mean that it's impossible to say that a language is more likely to have a dative than only an instrumental, for example. Though if by "dependency relation" you mean that "if A, then definitely also B", we just interpreted the question differently.