[lojban] Re: Denoting counterfactual sentences in Lojban?

[John E Clifford <lojban-out@lojban.org>]

Exppanding a bit

--- Robin Lee Powell
<rlpowell@digitalkingdom.org> wrote:
> On Fri, Apr 08, 2005 at 05:54:33PM -0400, Ben
> Goertzel wrote:
> > Hopefully someone on this list can clear
> things up for me.  I have
> > one significant question -- how to represent
> hypothetical
> > sentences 
> 
> da'i, or put it in an abstraction; all
> abstractions are technically
> non-veridical.

Well, {da'i} introduces the hypothesis and thus
starts a hypothetical context; it is a sort of
"if", in a word.  Thaat doesn't seem to be what
is wanted here and, indeed, it is a little hard
to call it hypothetical for just that reason. 
The suggestion to put it into abstract form, {nu}
or {du'u}, is a good one except that it is hard
to see what the appropriate sentences would be;
what selbri goes with the abstract sumti.  I
think the problem is that the sort of analysis
being proposed here just doesn't work very well
at the meaning level in Lojban (or English
either, for that matter).  To be sure, on a lot
of grammars, these underlying sentences would
have some grammatical existence but they would
never surface as sentences independently --
mainly just because they are subordinate in
non-assertive contexts and it is hard to assert
non-assertive sentences.  Maybe, add a "possibly"
in front and then say that so-and-so assumes or
knows "actually."  By the way, if Zhang knows
that someting is the case, the it is the case, so
for that you have an assertable sentence.  Of
course all of that is ultimately burtied in
another nonassertive context, so can't be
assetrted after all.
> > First, consider Exercise 7, part 2:
> > 
> > "Susan assumes that Zhang knows that Susan is
> late."
> 
> la .susan. sruma lo du'u la .zang. djuno lo
> du'u la .susan. lerci
> 
> > The translation given in the answer key is:
> > 
> > .i la suzyn. sruma lenu la jan. djuno lenu
> lerci fa lenu la suzyn.
> > klama
> 
> Whether to use nu or du'u is stylistic, for the
> most part, in this
> case.

Well, knowing a state of affairs doesn't make a
lot of sense, nor does assuming one.  I would
think {du'u} is pretty standard for both of
these.
 
> > Now, this is OK but personally I find it a
> bit annoying.  I found
> > myself wanting to do instead something like
> > 
> > .i lenu la suzyn. klama cu lerci     ("The
> event of Susan coming is late")
> > .i la jan jimpe go'i   ("Zhang knows the
> previous.")
> > .i la suzyn. sruma lenu go'i    ("Susan
> assumes the previous.")
> > 
> > or else replacing the last of the three
> sentences with
> 
> You're certainly welcome to if you like.  It
> seems obnoxiously
> verbose to me, however.

And, of course, says something different from the
original as presented.  Which is why you want
somrthing for marking non-assertive sentences
(not exactly hypotheticals).  
 
> > .i ra srumo lenu go'i (using "ra" to refer to
> "Susan", pretty
> > obviously in context)
> 
> Use "sy" to refer to Susan.
> 
> > However, I don't yet know how to mark the
> second utterance in this
> > chain as hypothetical, so that the listener
> knows I don't really
> > believe Zhang knows the previous, I'm just
> saying that Susan
> > assumes so.  IN other words, I want to say
> > 
> > .i la jan HYP jimpe go'i   ("Zhang knows the
> previous.")
> 
> da'i
> 
> Alos pe'i, ti'a, .ia.  Probably some others.

These are evidentials; they don't interfere with
the assertiveness of the sentence they occur in
but they indicate a measure of uncertainty (or
certainty) which opens an out if the sentence
proves to be false after all.  They are not quite
what is needed (but I am not sure there is
something for what is needed here).

> > I don't like
> > 
> > .i la lojban HYP mintu le glibau
> > 
> > because this is a posited equivalence between
> two entities of
> > different types, it seems semantically
> incorrect even though it
> > may (?) be syntactically allowable.
> 
> You mean because it equates a cmene and a
> lujvo?  It doesn't matter;
> they are both names, because of la.  It's the
> referents that matter.
> 
> gliban, btw.
> 
Or, for that matter, {la inglic} or with one or
more /i/ replaced by /e/.