[YG Conlang Archives] > [ceqli group] > messages [Date Index] [Thread Index] >
Rex:
[...]
#> Although 'Lojlan' seems verbose, it is not because of the obligatory
#> marking of grammatical categories. Rather, it is for three reasons.
#> First, missing out little words yields a structure with a meaning
#> other than what is intended. Second, and more crucially, concision
#> was simply not a design goal; although actual users are indubitably
#> guided in their usage by factors like syllable-count. Hence the
#> Lojlan design contains unexploited potential for abbreviatory
#> constructions. And thirdly, the phonology and the morphological
#> design means that the supply of short words and 'little'/function
#> words is very very limited, and now exhausted.
#
#Yes. Here's my thinking. In Loglan, if you say ciq stu, there is a
#precise, unambiguous meaning there. In Ceqli, not so. Ciq stu can mean
#heaps of things:
In Lojlan there is a precise, unambiguous & extremely vague meaning
there.
#Go ciq ke zi stu. Kyu zi ciq ke go stu? Da pa ciq ke go stu. etc. Any of
#these can be clear from the context. My basic point is that Loglan just
#can't let context modify the meaning. The unambiguity is a good thing, but
#I want to have it -optional-.
Lojlan has no trouble at all being vague and letting context clarify the
meaning -- most Lojban text is just like that. The problem with Lojlan
is not that you can't be vague and rely on context but that you can't
write in the telegraphic style of personal journals and of newspaper
headlines -- "man bites dog". It's easy to achieve vagueness but
hard to achieve brevity.
#> #Meanwhile, how can I best
#> #-retain- the rigor of Loglan while making it -optional-?
#>
#> If you did want to do that, then based on what I say above, I
#> would recommend taking predicate logic as a starting point,
#> as Loglan did, and, also as a starting point, an unambiguous
#> grammar (on which I could say more if asked). The unambiguity
#> of the grammar of Lojlan was added only as an afterthought, which
#> makes it a bit clunky; if it had been there from the start it could
#> be simpler. You could then either look to create new constructions
#> that abbreviate chunks of logical structure, or else take word
#> combinations that have no legal parse under the core grammar,
#> and use those for the 'rigorless' version of the language.
#
#Now, does predicated logic require the argument-places of Loglan, or have
#alternatives been developed? I seem to remember a set of case markers.
Predicate logic requires only that arguments be distinguished from one
another. Standard notation relies on ordering, but 'case markers' would
be equally okay. Exactly the same goes for natural language predicates,
as of course you know.
#Also, I was put off by the seemingly elegant, but operationally awkward,
#fact that 'mrenu' has to mean 'be a man,' and can't be used metaphorically,
#as in 'man the pumps.'
True, though it's easy to derive verbs meaning "beman (the pumps)" and
"hammer-use". By disallowing zero-derivation (i.e. change of sense without
change of form), ambiguity/homonymy is avoided.
#> But if I look at where Ceqli is now, its grammar is very similar to
#> English, but miles away from the 'logical' grammar I've been
#> describing, and hence my conviction that this sort of ultralogicality
#> is not something that Ceqli aspires to, because it conflicts
#> with far more important goals of the language.
#
#Maybe so. I'm hoping you or some other loglanist will help come up with a
#set of disambiguating particles that will at least move Ceqli in the
#-direction- of Loglan. You've noticed, I'm sure, that I have some
#elementary attempts at systematic parenthetization (the to...beto, etc.),
#and an as yet undeveloped set of t-words to correspond with the Loglan
#l-words (la, le, lo, lea, etc.).
Here's my advice:
1. Forget about logic, if it's not something you care deeply about. Logic
fans can go to Lojban.
2. Start with a simple grammar that follows these two principles:
A. Heads precede their dependents.
B. Dependents are never optional.
Those will pretty much guarantee you an unambiguous syntax that is
simple and easy to grasp. Then in places where you find it too
constraining -- e.g. where you want subjects to precede verbs, or
where you want complements to be optional, you can think of
devices, gizmos and wheezes to work them into the grammar
without creating ambiguity. Alternatively, you could have a
set of words that begin sentences or clauses and guarantee
that the rest of the sentence/clause is syntactically unambiguous,
and say that any other strings of words count as sentences that
may be syntactically unambiguous.
--And