Re: free word-order conlangs

---In conlang@yahoogroups.com, Patrick Littell <puchitao@...> wrote:
[snip]
>Yes, that's one of them; it is indeed Australian. This is a common
>sort of thing in Warlpiri's family, although I couldn't say if every
>Pama-Nyungan language has this feature, or whether this feature is
>widely-spread outside of the Pama-Nyungan family in other languages
>of Australia.
Thanks.

>>Some languages allow you to move clauses around within sentences,
>>but not to move phrases around within clauses.
>>Some languages allow you to move phrases around within clauses, but
>>not to move words around within phrases.
>>Some languages allow you to move clauses around within sentences >>and
also allow you to move phrases around within clauses, but do
>>not allow you to move phrases out of their "home" clauses within
>>the sentence.
>That's pretty much correct; what I said in my implicational hierarchy
>was a sort of oblique way of saying that. When we have the hierarchy
>w > x > y > z, we can either say "there exist languages with w that
>don't x, and with w and x that don't y, etc." or we can say "every
>language with x also ws, and every language with y also ws and xs,
>etc."  It comes to the same in the end.
But I was -- and still am, I guess -- confused about what you are
hypothesizing.

I came up with four statements that may be your hypothesis.

Let the features (there are five of them) be;
(A): You can move clauses around within sentences.
(B): You can move phrases around within clauses.
(C): You can move words around within phrases.
(D): You can move phrases _out_ of their "home" clauses within sentences.
(E): You can move words _out_ of their "home" phrases within clauses.

The two hypotheses I bet on were:
1. (D) implies both (A) and (B).
and
2. (E) implies both (B) and (C).

I thought you might be proposing one or both of those.

But I couldn't help wondering whether you were hypothesizing, instead, one
or both of
3. (B) implies (A).
or
4. (C) implies (B).

I wouldn't know how to bet on those.

Two hypotheses which I bet _against_ were;
5. "(A) and (B)" together imply (D).
and
6. "(B) and (C)" together imply (E).

>We'd have to be careful of the last one, though.
Yes, I see.

>Allowing a phrase to "runaway" from its "home" -- it's a good
>metaphor, btw
Thanks. I'm only responsible for the "home" part -- that is, I thought of
it without any input from the original inventor, if that wasn't me.
Somebody else thought of the "runaway" part.

>-- is different than the near-complete disregard for constituency
>we'd find in, say, Warlpiri.
Because they might not have "homes"?

But I'd still like to classify Warlpiri as having property (D) and/or
property (E) as above.  The problem, then, would be to re-phrase those
properties so that they still take in languages like Warlpiri.  (Or, of
course, a different solution would be, "don't do that; just give up on
Warlpiri".)

>Just running away from home isn't all that strange -- And gives us a
>few in English, Russian gives us even more freedom, etc. In these
>cases, the words still clearly have "homes", whereas we might
>stretch the metaphor to say that words in Warlpiri are basically
>nomadic.
I see.  Or at least I can now locate my blind spot. < :-) >

>>Consider it's _me_ you're replying to; have mercy on my
>>limitations, and please be explicit; _what_'s the motivation again?
>>That the most emphasized "thing", whether it is an immediate
>>constituent or something smaller (a constituent of a constituent) --
>>that is, whether it's a noun-phrase or a word within a noun-phrase
>>-- can be moved to the front (even if that means breaking up the
>>middle-sized constituent of which it is part)?
>
>Yup, the motivation would, in this case, be emphasis.  When we say
>"motivation" in this case it just means "why is the constituent
>moving from where we'd expect it?"  A modern Chomskian theory is
>going to require motivation for each transformation -- you can't
>just move things for no reason at all.
AFMCL there are going to be basically four reasons AFAICT so far.
Topics (may) "move" to the front; (they might have been there in the first
place);
Foci (may) "move" to the end; (this is focus-of-emphasis, not point-of-view
or perspective; and some sentences may have more than one such focus);
"Presentative" information (may) move to the end; (information the speaker
thinks the addressee is going to need to remember and keep in mind for the
upcoming discourse);
"Empathetic subjects" or "perspectival subjects" or "point-of-view
participants" (may) need to move close to the verb; (actually a given
clause may have a secondary one of these, which will then need to move
closer to the verb than any but the first one; and I'm toying with having
some clauses have a tertiary point-of-view).

>This is why there's so much talk about Warlpiri and its relatives. If
>we posit that sentences are base generated with constituents, Russian
>doesn't provide a problem, since that one movement is "motivated"...
>Warlpiri proposes a problem for this: if Warlpiri sentences are
>generated with constituents, then we have to come up with some reason
>why they scramble themselves all over the place.  Specifically, a
>reason for why does this word move there, and not just a blanked
>statement "They do so for informational-saliency reasons."
Don't some linguists think things just don't move -- that they stay where
they are born?

Even if the linguists themselves don't think so, don't some of their
grammars "think" so?  I've read (whether the writer was correct or not)
that the grammars used successfully in machine translations are likelier to
be based on "surface" structure than any idea of "movement" from some
generated "deep" structure.

>So some linguists will hold up Warlpiri as a counterexample to the
>idea that sentences are generated "underlyingly" with constituency.
I can see why.  I wonder whether it really is such a counterexample?

That statement is much more specific than the following one, which appears
ridiculous on the face of it.  (Not that the face of it is necessarily all
there is to it -- FAIK somebody who knows more than me has actually
successfully defended this idea.)

>That is, a theory that says "Warlpiri looks like English in
>underneath but all the words get scrambled"
That's the idea I said was ridiculous; the Warlpiri speakers generate their
sentences in "English" (except for the vocabulary), and then move the words
around.

>would have to abandon the idea that all movement happens for some
>reason.
Isn't there a way to generate Warlpiri without having to move anything
after the generation?  Or is that question beside-the-point?

> -- Pat
Thanks.
-----
eldin