[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >
'o~lelo 'o And:
> pc:
> > araizen@hidden.email writes:
> >
> > I once thought about implementing a "reverse polish" loglan
which
> > would be more or less isomorphic to lojban, i.e. all terminators
and
> > scope-marking would be implicit in the structure of the grammar
> > *without* using explicit words
> >
> > Can you really do this. Logic gets away with it by clear sentence
> > beginnings, HP by hitting the invisible "enter" key; how does a
> > language as complex as Lojban manage
>
> RP would be murder, I reckon: you'd not be able to parse until you
> hit the end of the sentence. Livagian, my engelang, uses polish
(i.e.
> head-first), and also allows no unmarked variation in the number and
> ordering of a word's dependents. The result is that it can be
unambiguously
> parsed incrementally with no lookahead and no backtracking.
Flexibility
> of constituent order is achieved by having devices that explicitly
> mark reorderings and changes to the number of a word's dependents.
I'm not quite sure what you mean by 'parse'. I thought about using
polish notation, and I don't think it really matters, but I decided on
reverse polish because that way you only need to keep a single stack
in mind, instead of a stack of arguments and a stack of operators (and
also I kind of like postpositions). Certainly, either way can easily
be transformed into the other.
The biggest problem is the variable number of sumti in a single bridi.
What I would do is group the arguments that form a terbri into a
single argument whose beginning is marked, and have selbri each act on
one such list. I use 'be' for this purpose, since it introduces a list
of sumti, and, since every selbri can pontentially have linkargs, it
is required with every selbri. (I had actually come up with a
simplified morphology and a way to move lojban words into the new
morphology, but here I'll just use the lojban words in new ways for
the sake of recognizability.) This word 'be' is a punctuation mark, I
guess, but I don't see any better way to do it, so I just keep the
punctuation to a minimum.
For example, 'mi prami do' would come out as 'be mi do prami'. The
'be' indicates that all following sumti are grouped together into a
single terbri (until you get to a selbri). 'mi prami do .i mi viska
do' would come out as 'be mi do prami be mi do viska'. (In order to do
tanru, you would have to use an explicit operator on two different
selbri.)
'le' turns the previous selbri into a sumti (which one can be marked
by 'ke'a' in the terbri if necessary). Sub-bridi, such as 'le nu' and
'le du'u' can be dealt with by simply putting the bridi in a the
correct place in the argument structure. For example:
mi prami le xrula
be mi be xrula le prami (the second 'be' is necessary to mark the
terbri of xrula, and to avoid claiming that I'm a flower).
mi djica le nu citka
be mi be citka djica
mi nelci le nu mi kansa do
be mi be mi do kansa nelci
le nu mi citka cu rinka le nu mi gleki
be be mi citka be mi gleki rinka
mi prami do mu'i le nu do prami mi
be mi do be do mi prami mu'i prami
That's the general idea. I suppose that the word order might wreak
havoc with English-speaking minds, but what's a loglan for? :-)
mu'o mi'e .adam.