Message-Id: <199602281858.NAA04552@locke.ccil.org>
Date:         Wed, 28 Feb 1996 11:03:39 -0500
Reply-To: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
From: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Subject:      Re: xelfanva and comments
To: John Cowan <cowan@LOCKE.CCIL.ORG>
Content-Length: 4287

>I wrote:
>>> .i va'o lenu cumki
>>>kei troci co jarco le djacu tubnu ki'u lenu se pilno gi'e troci
>>                                                     ^ku
>>> co mipri le li'avro ki'u ja'o na se pilno
>
>>>and wherever possible to expose the plumbing on the grounds
>>>that it is functional, and conceal the location of the departure gates,
>>>presumably on the grounds that they are not.
>
>Is the {ku} really needed there?  I wondered about that.  I thought I
>remembered someone saying that {gi'e} always works on the top level of the
>sentence; i.e. that you couldn't have  {lenu x gi'e y kei} because gi'e
>would automatically pop you out of the lenu clause.  Did I misunderstand?


NU/nu takes a "sentence" and GEkked and GIhEked constructs are subordinate
to sentence, therefore GIhE will not pop you out of a NU clause - only
something that cannot occur at that particuar point of a sentence.

>Lojbab replies:
>>I think you just missed a way to convey the complexity and the
>>parallelism:  termsets.
>
>I didn't even think of termsets.  They appear to be less awkward than what I
>was trying to do.

I think that complex sentencein translation are about the only places where
the parallelism of sentences makes sense.

>I do see you've made some better word choices than I had: zukte, jursa,
>terpli, tolja'o.

I was surprised at how easily they came to me too, since the last Lojban that
I did was at LogFest alst August.

But your word choices weren't bad - I just tend to be a perfectionist at
times %^)  And the chance to set an example for those who seem to think
that lujvo are second class words was certainly in mind as I tried to
write.  I like the subtety of meaning that comes from playing with lujvo.

> I don't understand why you put {tu'a} before {lenu lei
>farsni...}; it's already
>an abstraction.

Yes, but not the RIGHT abstraction.  This is the x2 of zukte, and therefore
an action (presumably taken by the agent).  The abstraction in the lenu clause
is not an agentive action, but is itself a result of such an action. Hence
you have an implicit secondlevel abstraction to bring about the result.
Maybe something like build/hang signs which point ... is the underlying
semantics.

But then one could say that the bare abstraction without the tu'a would
implicitly contain a "zu'e vo'a" simply because it is supposed to be the
action of vo'a, you copuld probably get away without the tu'a.  zukte is
one of the few gismu that has explicit sumti-raising built into the semantics.
But I would have made the "zu'e vo'a" explicit given that this is not a
well-known property, and those words would have been added with no direct
counterpart in the English original, which I find aesthetically displeasing
in a translation.

>How does {le teno'a} work inside a termset -- the outer bridi has several
>x3's, one
>per termset; does it refer to the one inside the termset where {teno'a} is
>used?  What if {le teno'a} were used outside a termset?

This may have been the first use of no'a other than in a refgrammar example
for all I know .  I was just winguing it, using my instincts for the language
on how >I< would want to interpret no'a in such a context.

In this case, I relied onthe fact that logical connectives can break out into
separate sentences, and if yoyu expanded this mess into separate sentences
then the proper place would be determined by the x3 of the immediately
localcomponent.  It is plausible to have it refer to the collected x3s
that are present, but I would have used vo'i for that.

The bottom line is that no'a is not used a lot, and I was trying to do the
one thing we do not have good Lojban devices for - sumti forward reference
(what I have called cataphora vice anaphora, to pc's humor).  I could have
been pedantic and exact with a negative subscript "ri", but the thought
turned my stomach in knots.  I am sure that someday someone will be forced to
such lengths to preserve the ordering of phrases in a translation.  Maybe I
am surprised that And Rosta hasn;t tried negative subscripts too.

Now free associating, I contemplate a negative subscript in the proposed
fuzzy logic convention for subscripting ja'a %^)  Anyone care to contemplate
negative truth values and negative ordinal members of a closed sequence?

lojbab