Re: Triggeriness ...

Andreas Johansson wrote:

> Perhaps I should consider the sparsity of answers to the questions in the
> below post as a sign that no-one feels like enlightening me,
Now,now, no need to get snippy :-)))))))))))))

but I chose to
> believe it just got lost between the list being held and the flurry of
other
> posts about trigger systems.
Bien sûr..............
>
> So, is the system sketched below a trigger system, and if no, why not,
<hopefully expert opinion>Yes, I think it is </hopefully expert opinion>

>and if
> yes, why couldn't we then subclassify trigger languages as nominativesque,
> ergativesque, and so on?
>
Perhaps I'm being obtuse-- don't these two terms refer only to how a
language handles A and P (or S and O)?  If you can trigger on ~focus
~subjectivize LOC et al. , where are you then? A LOC is neither an A nor a
P.  And presumably in a trigger conlang, you could trigger on any role you
wanted to (maybe within limits....)-- try GEN

I stole John's mango
John_TRIG stole-GEN mango-P I-A
(it seems possible, but I've no idea how it could be equivalently translated
into e.g. Engl., as we can paraphrase Pool-TRIG bathe-LOC I-A [pool swam-in
[by] me] =~ a pool is where I swam (?)---maybe 'John is the person whose
mango I stole'????)

In old TG, this would derive from 2 underlying S's, something like--:
S[I - steal - mango+S'[mango belong John"]S']S --so it's unlikely that GEN
is a valid role/valence of {steal}.

For brevity, I'll snip your ex.sentences, but it seems to me you want to
have special markings for S/A/P even when some other role is triggered.
This might be necessary to avoid ambiguity (in the pool, of "shark" and "I",
who killed whom?-- for clarity, actually you only need to mark one); but in
many other cases e.g.[inanim.object] that would be unnecessary).  Still it
could be done; there's nothing wrong with a little redundancy :-)

> > Also, since this is apparently NOT how a trigger language works...,
Well, it is, sort-of.........

> >...what would one call a language that DOES work like this, and are there
any?

A trigger language with obligatory case marking?? I don't know of any, but
maybe ANADEW.  (One could view Tag. ang, ng, sa etc. as case markers, but my
feeling is more that they simply mark subsidiary (non-triggered
~non-focused) elements. And those depend on the underlying valency of the
verb.  And some verbs are tricky-- we think of {swim} as mono-valent
(requires only A, LOC is optional) but it can be sort-or bi-valent in
"Ms.Ederle swam the English Channel" which can be passivized, but is "the
Engl.Channel" a P or a LOC? (IMO the active version seems to have a deleted
"across". )
{love} OTOH seems to require only A and P, _can't_ have a LOC (?*I love her
in the kitchen [but nowhere else???]), while {hit} requires A and P, and can
have INST and LOC.

Have you ever dipped into Filmore's "case grammar"-- ("The Case for Case" is
often anthologized, may even be online.)  I found it a welcome alternative
to Chomskyism, even though it never developed very far.

Hope I have made some sense.............

Replies