Re: [lojban] Re: Is Lojban a CFG? (was Re: [lojban-beginners] Re: Enumerating in Lojban)

["Jonathan Gibbons" <jonored@gmail.com>]

> 1.  "Not a CFG" != "not formalizable".  Lojban is still formalizable
> and easily machine parseable (witness the PEG formalism, for
> example).

My question being, why build in a constraint that makes it not a CFL,
with no significant benefit that I can see, which requires the use of
a more powerful formalism, that has been much less explored and has
fewer algorithms and a smaller codebase, to describe the language?

("Formalizable" is not exactly saying much about a language; I can
formally describe the language consisting of descriptions of turing
machine description and input pairs such that the turing machine will
halt given the input; on the other hand, it is provably impossible to
write a parser for that language, but that's not the main point of
your statement.)

> 2.  Elidable terminators seem to make intuitive sense to humans.

Actually, for my part, the aspect of them that I am suggesting being
different was very much /not/ intuitive. I expected them to behave in
a manner similar to parentheses; if they are there, then they force
the meaning to be a particular thing, if they aren't, then the meaning
may or may not be as intended, dependent on other rules involved. This
confused me for a very long time with the various, as I was making
statements that I was not sure if the terminator was neccessary or
not, and I expected to have it spit back something that I didn't mean
if I was wrong about needing them, and instead nothing/cryptic syntax
errors came out; so I spent quite a bit of time trying to figure out
what it was, and lack of an elidable terminator was generally the last
thing I thought of, because it was unintuitive to me to have a the
absence of what I saw as a disambiguation mechanism cause a string to
not parse at all.

> 3.  The lanugage would be *very* unwieldy as a CFG: way to many
> syllables would be expended in required required terminators.

I'm not suggesting dropping the concept; I'm suggesting making them
optional, and defining a rule or two on the semantic side of things to
disambiguate what would be ambiguous. As such, human use would be
almost exactly the same; it's just that if you elide a terminator
where one is not permitted to be elided now, you would  possibly be
saying something that is not what you intended to say, rather than a
decider for the language being required to reject it. Having a decider
for the language reject statements that a human would find the meaning
of unintuitive does not seem worth dropping being a CFL for.

-Jonathan