From: Logical Language Group <lojbab>
Message-Id: <199411102227.AA18599@access2.digex.net>
Subject: Cowan's summary: opacity and sumti-raising
Date: Thu, 10 Nov 1994 17:27:29 -0500 (EST)
Cc: lojbab@access.digex.net (Logical Language Group)
X-Mailer: ELM [version 2.4 PL24beta]
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length:       2830
Status: RO
X-From-Space-Date: Thu Nov 10 17:28:11 1994
X-From-Space-Address: lojbab

After some voice conversations with pc and lojbab, I think there is a fairly
clearcut resolution for the problems that have been agitating the List since
August, or whenever.  The short version is:  most of Jorge's points are
well-taken, and his views are for the most part sound; however, his actual
proposal ("xe'e") doesn't seem to be necessary.

This all began when Jorge objected to sentences like:

1)	mi djica lo ckafi .a lo tcati
	I desire some portion-of-coffee or some-portion-of tea.

As he rightly says, this form is equivalent to:

2)	mi djica lo ckafi .ija mi djica lo tcati
	I desire some coffee, or I desire some tea

is true, and I may truthfully say this if I desire coffee but not tea
(or vice versa).  When my interlocutor, wishing to be helpful, brings me
the wrong drink, I may refuse it, for if I want coffee but I don't want tea,
Examples 1 and 2 are still true statements, since (true OR false) = true.

The answer is that a desire for either coffee or tea (I don't care which)
is an inner-scope prenex; we may reword Examples 1 and 2 as:

3)	da poi ckafi gi'a tcati zo'u mi djica da
	there-is-some-X which is-coffee or is-tea : I desire X

whereas the more normal interpretation of "I want coffee or tea" is:

4)	mi djica le nu da poi ckafi gi'a tcati cu co'e
	I desire the event-of some-X which-is coffee or tea having-some-property

where "co'e" expands to something like "se ponse mi" = "is possessed by me".

This can be concisely rewritten as:

5)	mi djica tu'a lo ckafi .a lo tcati
	I desire some-abstraction-involving some coffee or some tea

pc has stated (and I believe he is correct) that all these opaque contexts
(seek, desire, need, etc.) always involve a hidden abstraction.  Some, like
"seek", always involve an abstraction; others, like "need" may sometimes
involve an object rather than an abstraction:  you may simply need that
there >be< an X, rather than needing to >do< something with X.  The
appropriate way, then, to get an opaque reading of a sumti is to either make
it explicitly an abstraction or to mark it with "tu'a", which creates a vague
abstraction from a concrete sumti.  In general, I believe that almost any
"concrete" place may be filled with an abstraction under the right circumstances
(I find it difficult to conceive of a process or a quality in the x1 of
"gerku", but who knows?").  This makes the careful use of "tu'a" even more
important, because sentences like:

6)	mi nitcu lo tanxe
	There is a box I need.

and

7)	mi nitcu tu'a lo tanxe
	I need a box (which may or may not exist; i.e. I need something
		with appropriately box-ish properties)

now mean different things.

I have swept under the rug so far any differences between "lo" and "da poi".
See part 2.

-- 
John Cowan		sharing account <lojbab@access.digex.net> for now
		e'osai ko sarji la lojban.