[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
From: pycyn@hidden.email Date: Tue, 7 Jan 2003 12:27:56 EST Subject: Re: [jboske] Monty's Unicorns, Fermat version To: lojbab@hidden.emailPutting myself back forty years to be in a Montague class is just the sort of thing I don't have time for now. In addition, I have said my piece numerous times on these issues -- with no visible effect then and not likely to have more now. But some thing popped into my head just reading this note through, so I'll pass them on for what they are worth.On {sisku}, I note that a correction that was approved ages ago -- a kludge but a useful one -- is not incorporated into the list: 1 seeks an instantiation of property 2 in set 3. We've always known -- and several times, I think, dealt with -- the fact that we don't look for properties but for their instantiations. Along that line, I recall (not a pramana, remember) that the quantifier problem was dealt with (as usual in intensions) by incorporation into the intension {leka pavyseljirna remei/romei}.[I've said a bunch of things about {ka} that mean that I am less than happy with this reading, but they don't come up here. Well, one does: {ka ce'u blanu} is a functon from individuals to colors -- whatever they are -- not from worlds and times to truth values, at least no directly. The latter is {du'u ce'u blanu}]The bit about a name having to have a property to be used comes from the fact that names are quantifiers (this cuts the grammar size roughly in half, eliminating a vast array of duplicates) and quantifiers are all restricted (second order relations between sets). Taking the properties to be a haeceity was a mistake I remember arguing with (probably) Gaifman back when I was studying to be a Nyayika and so a believer in visheshas. Even without vishesha, using this as haeceity seems to me a bad idea, since it makes transworld comparisons (ctfs like "If Socrates were a Seventeenth century Irish washerwoman") impossible to deal with naturally.Nick's run theough the technicals of Montague look OK, but Montague is easy to make plausible looking lines in that turn out to be either wrong or meaningless. And getting into the set where I can check is just what I can't do now.On the tense logic matter, strict tense logic would make {pu} and {ba} (and {ca} for what it is worth) particular quantifiers over past and future times, respectively. Thus, inside the scope of a negation they would be effectively universal. We did not go that way, taking them as pointing to fixed (if unidentified) times -- names rather than quantifiers, if you will. Or as giving them scope even broader than quantifiers. Or (I think this is the official line) as merely directions, not pointal at all (but still unaffected by negations). To do universals (and even explicit particulars, etc.) we use {roi} -- though I don't remember whether it is {pu noroi} or {noroi pu} for "never in the past."Default quantifiers were indeed originally intended Griceanly -- and in direct imitation of natural languages generally. I think that is right for external quantifiers; I am less sure about internal ones, which I would prefer just to have as optional. We usually want a particular part of the referent -- one member or several or all -- and in most languages in most contexts it is clear which without saying excplicitly. But we rarely care how many there are in all (more often with {le} than {lo} cases, in languages where these can be distinguished), so there is little Gricean mileage to be had from default internal quantifiers.
Back those damned unicorns. I still support the peculiar gadri (NOT any of the ones in standard -- nor, so far as I had seen when I took off, non-standrad [& and X] -- Lojban). As xorxes pointed out, what is involved is a buried quantifier -- one in another world if we must or one in what amounts to an equivalence -- though stronger than material. "I am looking for a unicorn" amounts to "I am on a quest which will be completed (/satisfied/relieved/...) just in case there is a unicorn I see (/capture /touch/...)" The fact that the critters involved here are regularly called "any one will do" shows that what is involved is purposive (do for what?) and intimately involved with notions like satisfaction. I would take this new gadri as an improper symbol, not translatable alone but pointing to longer expression that must be translated as a whole (that is, as a simple surface phenomenon from a very complex deep structure involving lexical replacements as well as syntactic ones).
-- lojbab lojbab@hidden.email Bob LeChevalier, President, The Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273 Artificial language Loglan/Lojban: http://www.lojban.org