[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >
On Thu, Oct 4, 2012 at 12:10 PM, Mike S. <maikxlx@gmail.com> wrote: > > I am starting to dislike the idea of explicit-sentences (let's just call > them parentheticals, or Ps for short) being part of the productions of the > basic words. That makes it impossible to think about simple things like > "coordinators" without keeping the optional parentheticals in the back of > one's mind. As seen, it also causes a problem with afterthoughts that > wouldn't occur otherwise. I think that parentheticals should be introduced > in the middle of the productions of constituents. This would add verbosity > to the rules, but it might also add clarity and common sense on balance. The Lojban BNF uses "#" to mark word classes (selma'o) that can be followed by indicators (which more or less correspond to our parentheticals). We could use something like that. It's a good idea to disallow them at the end of productions, they can be their own sentence after all. But I'm not sure about disallowing more than one parenthetical in a row, in fact I think you should have "P*" in all places where you have "P?". > I would write all nonterminal symbols in camelcase and all terminals > lowercase. I find it useful to mark selma'o, which then act as the syntactic terminals of a grammar that separates morphology from syntax proper. It's true that IllocutionaryOperator is easier than CA for someone just starting with the language, but once you know what CA is, I find the short form, and the differentiation of words from phrases, much more readable. Nothing stops us from having different variant notations as long as they all represent the same grammar, of course. > Sorry for all the long posts which amount to me thinking out loud on the > thread. Here's what I think looks pretty good, is grammatically > unambiguous, and includes all three types of tree-like constructs: > > Sentence := P | Formula > > P[arenthetical] := IllocutionaryOperator Interjection? Formula | > Interjection Why "Interjection" rather than "P"? I want to be able to say: ca'u co pnde'eka'a xa sma klme'eka "I wonder, my friend, where you are going." > Formula := CoreFormula | ForeTree P? Formula | Formula AfterTree This is ambiguous, because "he bbba hi" could be generated in two different ways. We are probably going to be assigning the same meaning to "(he bbba) hi" and "he (bba hi)", but still. To make it unambiguous we'd have to do something like: Formula := Formula-1 AfterTree* Formula-1 := CoreFormula | ForeTree P? Formula-1 which means all foretrees have to be generated before starting with the aftertrees. No parenthetical allowed before an aftertree? Formula := Formula-1 (P? AfterTree)* > ForeTree := he | Term P? ForeTree | Coordinator P? ForeTree P? ForeTree > > AfterTree := hi | Term P? AfterTree | Coordinator P? AfterTree P? > AfterTree > > CoreFormula := SimpleFormula | Term P? CoreFormula > > Term := UnaryOperator | ComplexBinder P? Formula | Coordinator P? Formula > > ComplexBinder := Binder | BinderCoordinator P? (UnaryOperator P?)* Binder > P? (UnaryOperator P?)* Binder > > BinderCoordinator := he'e | he'a | he'o I'm still pondering this, but my initial reaction is that I don't want a whole series of duplicate coordinators. > SimpleFormula := ( C C C* | q ( C? V? )* q ) Desinence > > UnaryOperator := ( b | f | d | m | n ) Desinence > > Binder := ( l | r | s | x | g ) Desinence "g" is a coordinator, not a binder, but it's to be replaced with jek anyway. > Coordinator := j Desinence > > IllocutionaryOperator := c Desinence > > Interjection := w Desinence > > > Desinence := V (k V)* > > V := ( a | e | i | o | u ) ( ‘ V )? > > C := b | c | d | f | g | j | k | l | m | n | p | r | s | t | v | x | z Other than the issue of binder coordinators, it seems basically sound to me. co ma'a xrxe