the problem of "go'i"
Jump to navigation
Jump to search
- tersmus: !broda( ); !broda( )
- zugz: I couldn't make any coherent sense of CLL's "replacement" mechanics for NA and tags
- zugz: I'd be interested if anyone can find sensible comprehensive rules incorporating them
- durka42: I'm sure {na broda .i ja'a go'i} is supposed to mean "!broda (); broda()"
- zugz: that's what CLL says, yes
- Ilmen: I'm not fond of this replacement system, but I guess it goes this way: when a tag is used with {go'i}, the interpreter will search the first occurrence of the same tag in the main-level of the target bridi, and replace it with the new tag-clause
- zugz: what about when {go'i} refers to connected bridi?
- durka42: but it doesn't work in practice?
- Ilmen: And na and ja'a are special, because they can replace each other
- zugz: ge na broda gi brode .i ja'a go'i
- Ilmen: zugz: I guess go'i refers to the underlying ge-bridi
- Ilmen: kaxyje'u fa lo du'u na broda kei lo du'u brode
- zugz: so do you replace in both? Or neither? Or in front?
- Ilmen: at least I can't see any other interpretation
- durka42: ie ja'a kaxyje'u
- Ilmen: zugz: as "na" isn't at top level, it can't be replaced I think
- durka42: that dilemma doesn't seem to depend on which "replacement" semantics are used
- durka42: because you can't use any GOhA to get at either bridi connected by {ge}, can you?
- Ilmen: That's like, {cumki fa lo nu broda} {na go'i} ---> the na negates cumki, and not broda
- zugz: I don't see why connectives should be different from tags for this
- Ilmen: durka42: the go'i series only target main-level bridi
- zugz: if a connective can mask inner tags and NA from being replaced, why not a tag?
- Ilmen: zugz: I think tags as well are masked if they're not at top-level
- durka42: tags are more like places than connectives ma'i mi
- zugz: do we really want {ba ja'a broda .i na go'i} -> {ba na broda} but {ba broda .i na go'i} -> {na ba broda}?
- zugz: durka42: they're scope-dependent, so in that sense they're like connectives
- durka42: have to think about that, I see advantages to both
- Ilmen: Indeed, in {ba na broda} {na go'i}, the top-level bridi should be "balvi fa lo nu na broda"
- Ilmen: so the interpretation "na go'i" -> "ba na broda" is weird
- durka42: but if {ca ko'a go'i} can replace the {ca} at whatever scope level it was in the bridi
- durka42: then {na go'i} doing the same thing seems to increase consistency
- zugz: you also have the problem that {ro da na go'i} wouldn't always be equivalent to {na ku su'o da go'i}
- durka42: this is a more serious problem :)
- zugz: durka42: yes, if you ditch NA replacement, you have to ditch tag replacement too
- durka42: but it would still be equivalent to {ro da na ku go'i}, yes?
- zugz: I don't know... depends how you read CLL, I suppose
- durka42: I don't want to drop tag replacement either...
- durka42: but I am a fan of quantifier rules that actually make sense
- zugz: yes, I think that's rather more important in a logical language than making common things easy to say
- zugz: not that the latter wouldn't be nice too
- durka42: what if there are multiple naku terms in a bridi
- Ilmen: The Japanese way of handling yes/no would have been easier (no replacement)
- Ilmen: no = jitfa fa lo nu go'i (and not na go'i)
- durka42: yeah, says that in the CLL too
- Ilmen: yes = jetnu (fa lo nu go'i)
- durka42: I think it's too late to change this though...
- durka42: I don't think it's too late to fix the {ro} issue
- durka42: so I guess I'm opening to fixing this issue as well :p
- durka42: I wish there could be a way to make it consistent without ditching tag replacement
- durka42: http://korp.alexburka.com/#?cqp=%5B(pos%20%3D%20%22PU%22%20%7C%20pos%20%3D%20%22BAI%22)%5D%20%5B%5D%20%5Bpos%20%3D%20%22GOhA%22%5D&stats_reduce=word&search_tab=2&search=cqp%7C%5B(pos%20%3D%20%22PU%22%20%7C%20pos%20%3D%20%22BAI%22)%5D%20%5Bpos%20%3D%20%22GOhA%22%5D
- durka42: one example I can think of is {A: xu do ba vimcu lo fesydakli / B: mi pu go'i}
- zugz: oh, I assumed only the same tag would get replaced
- durka42: maybe a bad example
- Ilmen: me too
- zugz: can you remember where this is in CLL?
- durka42: xu do cliva xeka'a lo trene
- durka42: ienai go'i xeka'a lo vinji
- Ilmen: zugz: besides, CLL claims that even NAhE can be replaced
- Ilmen: {mi no'e gleki} {mi je'a go'i}
- zugz: durka42: that might work anyway, depending on what the semantics of BAI like that are