[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >
On Fri, Aug 24, 2012 at 3:49 AM, Mike S. <maikxlx@gmail.com> wrote: > On Thu, Aug 23, 2012 at 7:40 PM, Jorge Llambías <jjllambias@hidden.email> > wrote: >> On Thu, Aug 23, 2012 at 3:16 PM, Mike S. <maikxlx@gmail.com> wrote: >> > >> > la xrma le sme je strafe bjrafe >> > "The horse is swiftly running." >> > >> > This is probably going to be common enough of a construction that we'll >> > want a shortcut. I propose a new binary, "joy": >> > >> > la xrma joy bjra stra >> > "A/the horse in the same event runs and is fast." >> > "The horse runs fast." >> > >> > ... in which some variable is bound and understood to be f-suffixed to >> > "bjra" and "stra". This variable could be anonymous, or better yet, >> > assign >> > it as "oy" to match "joy" (understood as bound as "loy smoy" until >> > assigned >> > again) for possible use elsewhere. >> >> I like it, but I'd use "ju" which is available. If it turns out it >> would be useful to have the vowel be a variable rather than there >> purely for morphological reasons, I would say use a different >> consonant as a new binding operator. > > Okay. Sounds good. We don't quite have a proper definition for "ju" though. ju bra cre <-> li smi je brafi crefi is fine, but it only deals with how ju works with two predicates. When joining two general formulas things get more complicated because -fi isn't directly attached to a formula. Every formula ends with a predicate though, so is it enough to say that the -fi provided by ju is attached to the final predicate of the formula? Is it just: ju "formula1" "formula2" <-> la sma je "formula1"-fa "formula2"-fa ? Probably not. > [...] So the rule probably should be > that every "ju" f-binds all predicates anywhere in its scope, except for > those in "ju" child trees. I don't know if that works either. You could have lots of things going on inside there and we probably don't want all subordinate predicates to be about the same event. I think what we need is for the resulting equivalent predicate to take the -fa, but I'm not sure how to formulate that. > Probably what we really need to do is to start writing down every > interpretation rule - one for each unique syntactic operation, and > particularly one for each binary operator. Also, what to do about things > like free variables in predicates. We have some of these already. Most > likely one rule will be that f-unbound predicates look up the tree for the > nearest "ju" and use the event-variable it is binding; otherwise, some > default. I'm thinking that at least predicates used in the restrictions of quantifiers will have to be excluded. > The generative rules get us the sentence but the interpretation rules get > us the meaning. It will get complicated, but since this grammar is so > miniscule, I am actually thinking it is possible. I think it's possible, but some ingenuity will be needed to get the rule right. mu'o mi'e xorxes