[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >
Jorge Llamb�as, On 02/10/2012 23:22:
On Tue, Oct 2, 2012 at 6:40 PM, And Rosta<and.rosta@hidden.email> wrote:Jorge Llamb�as, On 01/10/2012 23:38:"sa R P" is truth functionally equivalent to "sa sma je R P", and I see no reason to exclude from these the contrived cases where R (or P) don't contain a free a.It's at least as easy to exclude them as to include them. (If necessary they could be reincluded by special rules that make them meaningful and somehow useful.)As I see it, there's nothing special we need do to include them, because the basic syntax already generates them
This is what I think is untrue. Whether or not the grammar generates them depends on the rules you specify for syntactic binding relations. Those are the rules I was talking about.
and there's a natural extension for their meaning from their ordinary meaning. To exclude them you need to add a rule to specifically exclude them, so it does involve a certain (even if minimal) complication.
I've demonstrated that you don't need a rule to specifically exclude them. The simplest formulation of Xorban syntax that I can currently think of does exclude them.
The approach I'd favour is to make syntactic binding obligatory.Me too, but what exactly counts as syntactic binding?
A long-distance relationship whose presence (or absence) is signalled inflectionally by sameness of vowels and with a range of semantic interpretations (generally, but not exclusively, quantifier--variable binding).
For me it's something like this: "A binding operator removes any free occurrences of its variable from its complement formulas."
I don't understand "removes". And what is "its variable"? "Its variable" is the variable it is linked to by syntactic binding. (And changing to "A binding operator removes from its complement formula any free occurrences of the variable it syntactically-binds" doesn't make sense -- if it binds them, then they're not free.)
The special cases in which the complement formulas happen to have no free occurrences of the variable are just that, special degenerate cases. The resulting formula will have no free occurrences of the variable, which is exactly as it should be.
I don't understand this, but maybe if you have another go at explaining it I will.
Structures without the appropriate inflections for the binding relations are either ungrammatical or rescued by special rules adding extra (phonologically null) structure sufficient to provide the binding relations.Formulas with free variables need to be rescued somehow because otherwise they can't have meaning. They need an implicit binder.
By "free", you mean "without an explicit (phonologically overt) binder", rather than "without a binder (of any sort, explicit or implicit)", right? There's no point rescuing them and making them have meaning unless that meaning is somehow useful. John WC's idea of allowing the implicit quantifier to be of any sort strikes me as potentially worthwhile, but I think a positive case for rescuing them needs to be made.
Binders whose one or both complements happen to have no free occurrence of the variable they bind need no special action, because they already have a natural meaning.
Right. But as I tried to explain in my last message, if quantifiers don't have to be syntactic binders, you still need -- slightly more complicated -- rules to prohibit sameness of vowel between the quantifier and the variables it doesn't bind. That is, for a structure expressing "Ex Fy Gy", the phonological form "sa bcda fgha" is ungrammatical: the phonological form "sa bcda fgha" is licit iff it corresponds to a syntactic structure in which there is a syntactic binding relation between the quantifier and the argument-markers.
The alternative is to make all binding relations optional, But how would that be formalized? You'd need to define a relation 'potential binding relation' such that if X and Y are in a potential binding relation then X and Y inflect with the same vowel iff one binds the other. This strikes me as unproblematic, but nevertheless as more complicated than the alternative.I think binders always do the same to their complements: they bind any free occurrence of their variable. Anything else strikes me as more complicated.
To me that is bordering on gibberish, so I'm fairly sure we're not understanding each other. I can't work out what's at the root of our communication failure. Would it help if you provided a list of the syntactic rules you think are needed? In a couple of messages I've listed the rules I think are necessary (comprising a reformulation of some of the syntactic rules among the various rules listed on the Xorban basic grammar page, plus a formulation of candidate rules for syntactic binding). I should also add that if there are terminological difficulties, I am happy to be flexible about terminology. For example, if "syntactic binding" is confusing because it's hard to keep it distinct from "semantic binding", I'm happy to adopt a different term, such as, say, "govern" rather than "bind". --And.