joiku

From Lojban
Revision as of 08:44, 6 June 2014 by Gleki (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Why doesn't "le broda joi le brode" parse?

  • joiks are both sumti and tanru connectives. Thus, the parser gets to broda joi and expects another brivla afterwards, and when it encounters le, reports an error. The solution is to explicitly terminate le broda with ku, which in this case is not elidable.
  • do skudji lu le broda ku joi le brode li'u .e lu le broda joi brode li'u
  • The longer answer to this is that Lojban is defined by a LALR(1) grammar, so that any ambiguities have to be resolved with at most one token of lookahead. le broda joi le brode has to be disambiguated as early as broda: it has to know whether the selbri is over by the time it hits joi, or will be continued through joi. To resolve this, though, it would have to look past joi, at le. This means two tokens of lookahead, which is thus disallowed. (Yes, it's counterintuitive, but that's what you get for using YACC to define a human language.) For further discussion, see:
    • wiw.org/~jkominek/lojban/9104/msg00011.html (where Guy Steele points out issues with YACC);
    • and for amusement value, [1] (where Nick Nicholas first discovered this issue to his acute indignation).
      • Well, it's counterintuitive to English speakers, but someone with a lot of practice in Lojban (a native speaker?) might find it clearer. For example, take a phrase like:
        le broda nu brode joi le brodi
    • It starts to get a little tricky even for a human to quickly pick out whether the joi attaches to the brode, the nu brode or the whole le-sumti when there's no ku before the joi.
    • Lojban really likes terminators, and so if you find adding an extra one or two to be counterintuitive, you're probably going to have a hard time. :-)
    • Adam Raizen:
      (I also had a hard time accepting this, until I came across something like the above example, which made me do a double-take and rethink how counterintuitive this situation really is.)
  • Lojban does not like terminators, it only puts up with them when they are necessary. One mark of good style is phrasing your sentences in such a way that the use of terminators in minimized. Unfortunately, this kujoi case is one of the unavoidables. It is about the only time that the terminator ku needs to be used.
  • Adam:
    Terminators are such an integral part of Lojban that it shouldn't seem counterintuitive to a good speaker to add them. If the ku of le broda ku joi le brode is counterintuitive, what makes the ku'o in mi viska le nanmu poi slabu mi ku'o bu'u le panka any more intuitive? Personally, I think that having to rearrange a sentence in order to be able to elide terminators is at least as counterintuitive as saying the terminators themselves.
    • Nick Nicholas:
      • shrug* Adding ku before joi in this context has become a reflex action for me anyway. Just because I think something's wrong doesn't mean I won't do it :-) . Very good example, btw.
      • xod:
        So what would that parse as? I assume, since the second argument of joi is clearly a sumti, then the 1st one must be too, which means the proper interpretation should be le broda nu brode keiku joi le brodi
        • nitcion:
          You're doing more than one token of lookahead, I'm afraid. There is no reason for broda nu brode joi le brodi to be any more grammatical than le brode joi le brodi; the parser still doesn't know whether the selbri (now embedded in a nu-clause) is over by the time it hits joi.
  • After appearance of camxes it parses so the rest of this article is obsolete.
    • ...I suggest not to mislead people who are looking for official Lojban, whether one likes official Lojban or not, or considers it broken or not... mu'o mi'e .iesk. 21:42, 22 oktobero 2013 (UTC)
      • That's right. The official parser is written in YACC that requires joi ku. Although, camxes parser written by Robin Lee Powell and la .xorxes. allows to omit ku no one ever declared camxes official. mu'o mi'e la gleki