Re: Core case roles

En réponse à "Thomas R. Wier" <trwier@...>:

>
> Unfortunately, there are many, many languages that structurally
> distinguish the behavior of forces and that of agents.
So what? Many languages distinguish themes and patients structurally, but that
doesn't mean they cannot be unified at some lower level, as DeLancey's theory
does (and with compelling evidence that it's a correct way to do it). The whole
point here is not that some languages make some segregations (you can always
add structural segregations based on the semantics of the nouns), but that some
don't, which would be pretty strange if the segregation existed at the lowest
level.

  Take Lakhota,
> for example (<_n> refers to nasalization of vowels, and <Z> [Z]):
>
>  (1) Fred / Tuwá / Hok?ila wa_n (ix?e wa_n ú_n) oZa_nZa_nglepi ki
>               someone  boy    a     rock a    with window         the
>
>         ka-blétShe        / *wo-blétShe.
>         by.striking-break    action.from.a.distance-break
>         "Fred/someone/a boy broke the window (with a rock)."
>  (2) *ix?e wa_n oZa_nZa_nglepi ki  ka-blétShe
>       rock  a    window         the by.striking-break
>       *"The rock broke the window"
>  (3) (ix?é  wa_n ú_n) oZa_nZa_nglepi ki  ka-bletSha-pi
>       rock  a    with window         the by.striking-break-3Pl
>       "They [unspecified] broke the window."
>
> It is clear from examples (1) and (3) that human beings can be the
> subjects of transitive verbs, but inanimate objects cannot be so,
> after (2).
So what? It just means that in this language transitive verbs accept only
certain nouns as subjects and not others. It doesn't say anything about the
semantic roles. It's indeed another subject altogether: verb semantics (not
that they are completely independent subjects of course).

  There are many such languages, and the fact that many
> languages distinguish morphologically between an instrumental and
> an ergative case, while others do not, suggests that despite the
> similarities there is a genuine reason for distinguishing them
> semantically.
>
Not really. If you are interested into defining the semantic *core* cases, the
basic distinctions that are found cross-linguistically, then you cannot
separate instrumental and agent for the simple reason that when they can never
appear *both* as core cases (when both appear, at least one is always oblique),
and there are plenty of cases cross-linguistically where they are not marked
differently when they appear as *core* cases (the case "I broke the window"-
"the stone (that I threw) broke the window" is just one of them). Just use
Occam's Razor here. Why add unnecessary distinctions when you can explain the
facts as accurately (and even more accurately) and predict them with less
distinctions? DeLancey defines Agent as "the external cause of the change of
state (or Location) of the Theme" (thus again in relationship with the Theme
and Location, rather than extra-linguistically). As such, an instrument used as
*core argument* is an Agent all right. Afterwards, different languages can
treat different subsets of this category "Agent" differently, depending on noun
and verb semantics, but it doesn't say anything about the core thematic
relations, or core semantic roles.

Christophe.

http://rainbow.conlang.free.fr

Take your life as a movie: do not let anybody else play the leading role.