[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >
Jorge Llamb�as, On 02/09/2012 23:04:
On Sun, Sep 2, 2012 at 4:32 PM, And Rosta<and.rosta@hidden.email> wrote:Yes, you're right, I don't like your donkey rendition and nor can I bear the implicit binding rule. I would instead have a rule that unbound variables (others than those defined as taking implicit binding) force a word order reversal in which quantifiers follow predicates: la bcda fghi lu jklu mnpukika li = la bcda li fghi lu jklu mnpukikaShouldn't the first one then have been: la bcda fghi mnpukika li jklu lu Otherwise, why wouldn't "lu" try to bind to the left? (and fail, because there's only one formula there)
Yes, but the idea is that once "lu" has tried and failed to bind leftwards, it then has no option but to bind to the right. The rules are: Every word tries first to bond syntactically with words preceding it; if it fails, then it bonds syntactically with words following it. I haven't thought about something like "bcda la fgha", i.e. whether la can have one complement preceding and the other following, or whether both complements must either both follow or both precede.
It would be nightmarish to write a formal grammar with that rule, I would think.
Not if you choose the right formalism. Take the kind of rules you have so far, and strip away the element of linear precedence, leaving purely hierarchical structure. Then add the rule I've outlined above. Bob's your uncle.
But I'd be fine with *a specific series of variables*, e.g. V'u, preserving the same restriction when subsequently unbound.But then the speaker has to plan ahead.
And the hearer is forewarned to make the necessary effort to remember.
The idea here is that this is for afterthought use.
First of all, I think that a loglang demand of its speaker a fair bit of forward planning. Second, I think afterthought anaphoric relations should not rely on memory of preceding phonological form or completed syntactic structure, i.e. should not rely on stuff the mind forgets as soon as it's done with it.
Maybe the specific series of variables could be the single Vs, the ones used most often. This would ensure that memory requirements are kept low, because if you start using more (and longer) variables you can't rely on the listener keeping track of their restrictions.
I'd have used the single Vs, or at least one of them, for the "o'e" variable, tho I suppose that still leaves four for the afterthought. But for the reasons given above, I'm opposed to the afterthought scheme.
Another is that I can't work out how to say "If anybody sees anything, they'll probably eat it". Well, it can be done as:Why was it that this one didn't work: ra prna re sme ja na vskake li fi ctkake lkniI meant a reading "if somebody sees something", where there are existential quantifiers within the protasis. Your version (a valid reading but not the one I intended) quantifies only over actual people and things. Seeing as you haven't come up with something better, here's how I'd do donkey structures: la fa si frmri su xslu je pnsiku va drxiku lo msto dnksntncako "most is the propertion of states of affairs in which "si frmri su dnku je pnsiku nu drxiku" is the case that are states of affairs in which "si frmri su dnku je pnsiku ne drxiku" is the case" where vV is a bound unary operator whose sole function (but a vital and necessary one) is to do donkey sentences.How about this: la frmra le xsle li fi pnsake lo mjrtoki fo drxake Farmers, donkeys, states of affairs in which they own them, the majority of them are states of affairs in which they beat them. This one doesn't have any implicit binding, and for "mjrtoki" to make sense it must mean that the state of affairs of owning a donkey is identified with the corresponding state of affairs of beating it. Then the donkey(s) beaten must be the same one(s) owned, as they are in the same state of affairs.
I think this is acceptable as a Xorbanish solution. My vV solution is an economical way of expressing exactly the desried meaning, but it's probably a bit too baroque for Xorban. (Or is it? Actually, no, I think it's pretty neat. I don't think I'll withdraw the proposal yet.)
And for your other sentence: la fa vskeki lo fo ctkeki lknoka Someone seeing something, them eating it, the latter is likely if the former is the case. This doesn't involve a "re prne ri smi" quantification, just "le prne li smi".
le prne li smi la fa vskeki lo fo ctkeki lknoka OK. Good. I'm satisfied.
ma'a xrxe
O very neat! ma'a ndi [my Italian name, short for _Andino_]