[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
Jordan: > On Sat, Dec 21, 2002 at 08:34:19PM -0000, And Rosta wrote: > > Jordan: > > > On Wed, Dec 18, 2002 at 12:21:09PM -0500, John Cowan wrote: > > > > And Rosta scripsit: > > > > > > > > > Fair enough. But all the parser needs to know is the selmaho, no? > > > > > > > > Well, in one sense. But the whole idea of *elidable* terminators is > > > > that they can be elided because one (a parser or a human being) can > > > > reconstruct the unique (sequence of) terminators that are missing > > > > If there is more than one cmavo in an elidable-terminator selma'o, this > > > > is not possible: does > > > > > > > > le broda [] cu brode > > > > > > > > mean > > > > > > > > le broda ku cu brode > > > > > > > > or > > > > > > > > le broda goi'a cu brode > > > > > > Saying this is "impossible" is a bit of a stretch > > > > > > More accurately, it would suck for the parser to need to inspect > > > which terminators were specified instead of elided and then examine > > > which token it was so it can decide if ko'a changed or whatever > > > other foo And wants to pack into the things > > > > But the current parser doesn't do things like anaphora, logical > > structure, and so forth, does it? It ignores semantics, and simply > > pronounces whether the string is syntactically well-formed or > > not > > What's your point? My point is that I don't see why the parser would have to inspect which terminators were specified instead of elided and then examine which token it was so it can decide if ko'a changed or whatever other foo And wants to pack into the things. It seems to me that allowing multiple membership in a terminator selmaho does not oblige the parser to take on the task of constructing logical form. If you weren't saying otherwise, then I don't know why you said "more accurately, it would suck for the parser to...". What was the relevance of saying that? --And.