[lojban-beginners] Re: lojban qua lingua franca

[Andrew Archibald <archibal@math.mcgill.ca>]

On Sun, Mar 28, 2004 at 09:03:55PM +0100, GREGORY DYKE wrote:

> > Curiouser and curiouser said 
> > 
> > Explain to me slowly the difference between to grasp and to learn 
> > from the point of view of lojban grammar.
> 
> (btw, there was "t" too many in my previous mail... I meant that it's not as easy to grasp as it is to learn once the basics are grasped)

Hmm.  Perhaps you were using "grasp" to mean "understand the
underlying principles of the grammar" and "learn" to mean learning the
details of the grammar?  Of course, one also has to learn the gismu,
which takes time.

> Well... this whole place-structure business seemed a bit wierd to me at first. Especially as, when I learnt lojban, it was with an ancient series of lessons written by Athelstan, which, once the place-structure was explained, spent some effort explaining the SE cmavo along with how all the zo'e could be removed... 

Hmm.  I didn't find the place structure too difficult; in principle,
it's just fill in the places from left to right, plus a list of
shortcuts and transformations.  You can drop trailing {zo'e}.  You can
construct new selbri with arbitrarily permuted places using sequences
of {se} {te} {xe} etc.  And if you want to explicitly label which
place a thing goes in, you can use {fa} {fe} {fi} etc. 

What I find difficult is grasping the grouping operators.  I very
often find myself wanting simple general-purpose parentheses.  For
example, to produce something like "the yogic flyer" you have to use
{be}: {le volfli be le makfioga be'o}.  At least, I think so; I can't
find any such usage in the CLL but I can't find any other way to do
this and people have claimed it's right.  

Nobody has really explained it like this so I'm not sure, but it seems
like the usual way to attach things in lojban is

thing1 (special purpose connective) thing2 (special purpose
continuation) thing3 ... (special purpose terminator)

where the terminator can sometimes be eliminated "where this does not
involve ambiguity".  (This last condition always implies significant
computation or significant memory to verify).

But then there is {cu}, which seems to supply terminators for all
(most?) constructions at once.

And some operators have the form

(operator) text (special purpose terminator)

(This is my impression of how one tells the scope of {nu}, for example.)
In this case, the operator has some "natural extent" after which it
stops applying (a bridi in this case, sometimes a selbri); if you want
to end it early you can supply its terminator, but if you want to extend
it you must use some sort of grouping operator with tighter binding.

Some operators (such as {na'e}) support a variant on this: it can be
used as {na'e ke} ... {ke'e} or just as {na'e}, which I think binds
more tightly than {na'e ke}. 

I can't find any explanation at all for terminators or precedence for
zo'u.  I assume it applies to a single bridi, and to terminate it you
terminate the bridi?

Finally, there are various words that can be thrown in more-or-less at
random; for example, you can introduce attitudinals at all sorts of
places in the sentence, and it's not very clear to me how big a piece
of the sentence they're supposed to apply to (perhaps the last word
unless it's a terminator, in which case to the whole thing
terminated?)

I suspect I am not the only one who finds these grouping rules
complicated; the CLL contains many sentences like:

#Note the importance of using ``kei'' after ``su'u'' when the x2 of
#``su'u'' (or any other abstractor) is being specified; otherwise, the
#``be lo'' ends up inside the abstraction bridi.

By contrast, the meanings of the various constructs are not much
harder to learn than the meanings of gismu.


I must say, it's deceptive to claim that the number of gismu (about
1300?) is the size of the vocabulary to learn.  One must learn most of
these, some additional lujvo, and also many hundred (thousand?) cmavo,
along with any special syntax associated with them.  For example the
three different quoting operators all have different syntax, all
different from the usual lojban grouping.  {na} has special syntax (it
goes before the selbri and affects the whole bridi) which is not
entirely specified (how does it interact with tenses? it may or may
not have a different meaning depending on whether it's before or after
other just-before-the-bridi modifiers).  And {bo}, in addition to
binding more tightly, reverses the usual left-binding rule.



My impression is that there are too many special-case grammatical
objects, that is, that the syntax is too closely linked to the
semantics for the language to be as simple to learn as it could be.  I
think most cmavo could be given gismu-like standardized grammar (in
fact, exactly gismu-like in that they would have place structures and
be used in the same way) apart from a short list of purelsy structural
cmavo to indicate grouping.  But probably such a simple language would
be cumbersome to speak; in any case, since I do not speak lojban well,
it's foolish for me to think about improving on it. 


Andrew