Re: Underspecified verbs?
|From:||jesse stephen bangs <jaspax@...>|
|Date:||Saturday, May 19, 2001, 6:01|
daniel andreasson sikayal:
> > Intent. Cause. Part.
> > Actor + + +
> > Agent - + +
> > Participant - - +
> > Origin - + -
> > Causant + + -
> > Object - - -
> Very nice system! I looks like an active system, but with the
> semantics generalized to all the arguments, and not just
> intransitive verbs.
Thank you very much! Thanks also to all of the other people who gave
their comments. This system first showed up as my first attempt at an
"active language," but this was before I really knew what an active
language was, and so I wound up with what you see above. Not exactly, but
the basic idea was there.
I like the idea that this system is fluid and perhaps slightly unnatural,
because I'm actually using this system for the very ancient predecessor of
Yivríndil. The advantage of this is that the system I've described could
evolve into just about any imaginable syntactic form. In the Yivríndil
language group the system became predominantly Nom-Acc, but with some
In Old Yivríndil there were four core cases, which were descended from the
original system in this way (not switching Actor and Agent for now,
though I probably will later):
Actor --> Nominative
Object --> Accusative
Agent --> Instrumental
Participant --> Dative
However, the valency for active verbs was different than for stative
verbs--the subject/object of an active transitive verb were
Nominative/Accusative, while for stative verbs they were
Instrumental/Dative. Thus, the system was simultaneously active and
nominative! This made at least a little sense diachronically--originally
most active verbs like "run" or "take" would take [+intent] subjects,
which would correspond to Nominative in Old Yivríndil. Stative verbs like
"love" would often be [-intent], and so take Agent as the subject, which
would be Instrumental in OY.
There were other quirks of OY syntax, as well. All passive verbs took
Intrumental/Accusative subjects and objects, regardless of their original
valency. There were also trivalent verbs which usually took Nominative,
Accusative, and Dative. Fortunately, all of this confusion was lost in
the transition to current Yivríndil, wherein all verbs take
Nominative/Accusative in active and passive voices. The only remaining
fragment of the OY confusion is that stative verbs may still optionally
take their objects in the Dative case--but that's all.
> [+event] means that the predicate describes an event as opposed
> to a state. [+P/E/I] means that the predicate is performed, effected
> or instigated in some way (but not necessarily controlled) and
> [+control] means that the predicate is controlled. Compare "hiccup"
> and "run" to see the difference between P/I/E and CONTROL.
Makes a lot of sense. I understand this system, too, and it's also
elegant in it's own way.
> And then you can introduce new parameters such as "significant
> affectedness" and "empathy" [as in Eastern Pomo :) ].
Which reminds me a lot of the parameter "Change of State" that I
originally used. Then I realized it wasn't doing anything other than
marking Objects as [+COS], and so I got rid of it as superfluous.
> Anyway... I suppose all this falls under the category "other
> comments" :)
> I haven't seen a natlang doing it exactly as you have but that
> doesn't mean it's implausible. I find it both plausible and very
> elegant. Very good work!
<<sigh>>. If only the rest of the language were at all understandable or
usable right now. Since it's the ancestor of Yivríndil and Yivríndil was
made first, I have to reconstruct all sorts of things and it gets very
complicated. But it's fun.
Jesse S. Bangs email@example.com
"If you look at a thing nine hundred and ninety-nine times, you are
perfectly safe; if you look at it the thousandth time, you are in
frightful danger of seeing it for the first time."