Andreas Johansson wrote.
>Reminds me ... in a thread on the case system on one of my langs* someone
>suggested that there might be languages that has one case of intransitive
>subject (S), and another that covers both transitive subject (A) and
>transitive object (P). That seems Truly Weird (TM) to me, so does anybody
>know of a natlang that DOES work that way?

I believe that in *Describing Morphosyntax*, it was said that there was
only one language known to have an S vs. A/P distinction, and that it
was unstable even in that one. Essentially, the reason for this is that
such a distinction is, as you said, weird.

Basically, there are two functions of constituent case marking. The
first is discriminatory; that is, it helps distinguish between different
cases where otherwise they would be ambiguous. Since S and A will never
appear in the same clause, and S and P will never appear in the same
clause, there is no need for a distinction there; meanwhile, since A and
P *will* appear in the same clause, that distinction *would* be useful.

The second function is identifying, or uniting elements with similar
functions. S and A have the common trait of agentivity--if a clause has
an agent, it will be one of those two. Ther also share the trait of
topicality, meaning they're the roles most likely to be the topic of the

S and P, on the other hand, have the semantic similarity of state
change--if a clause has a patient, it will be either S or P. They are
also the who roles in which "new" participants appear most often.

So, S and A have certain traits in common, as well as S and P, meaning
it's likely that S will be lumped in with one of the two. A and P,
however, have not much at all in common, making them unlikely to be
lumped together.

I hope all this has been enlightening rather than confusing....

- Ian Maxwell
CU v1.1 !lm+ cN:R:N:H a+ y n19:1
!B* A+ E--- L+ N1<2 Im k-- ia@:@
p@ s@ m- o P-- S--- Ingyrric