From LOJBAN@CUVMB.BITNET Sat Mar 6 22:46:13 2010 Reply-To: Logical Language Group Sender: Lojban list Date: Tue Dec 19 07:19:45 1995 From: Logical Language Group Subject: Re: la lojban zasni X-To: sbelknap@UIC.EDU X-cc: lojban@cuvmb.cc.columbia.edu To: John Cowan Status: OR X-From-Space-Date: Tue Dec 19 07:19:45 1995 X-From-Space-Address: LOJBAN%CUVMB.BITNET@UBVM.CC.BUFFALO.EDU Message-ID: >>> 2. As part of the new version release, assure that there is a well-defined, >>> nonambiguous translation algorithm from la papinomoi lojban -> la papipamoi >>> lojban (or whatever). Thus all well-formed extant texts can be translated. >> >>I wonder how you would handle "changes" such as the normalizing of the >>use of {ri} in {le re nanmu cu prami le ri speni}. I expect most of the >>evolution of the language to be of this sort, getting a clearer >>understanding of already parser-grammatical Lojban. If rules of >>interpretation vary somewhat, it will be extremely hard, if at all >>possible, to "translate" from old texts. >--More-- > >I think perhaps the parser is less useful than it might be, if it allows >"legal" but as yet unexplored parts of the language to pass through. >Perhaps a second pass parser could label something like the use of {ri} as >*experimental* until it was well worked out. (I am *not* suggesting the >parser check for context!) I'm probably missing some of the context of this discussion, drowning back in last month's onslaught, but the issues that the parser does NOT resolve, which I think are being referred to here, are indeed those that require context checking. The rules for interpreting "ri" are almost certainly to at least a small extent context dependent, though one might be able to devise algorithmic rules for it. But those rules are set only in jello and not in concrete, as we have not tested all plausible varieties of backreference with "ri" (and may never actually do so for ALL variations). Some undecided loose ends include whether "ri" can refer to and pick up the value of an explicit "zo'e" place. I think it should, but since we have a standard that says that "ri" does not pick up the value of any monosyllable fixed analphora like mi/do or ko'a, there would be a conflict if an explicit zo'e stood for one of these values. The purpose of the parser is simply to identify ungrammatical text 9and possibly to say where it is ungrammatical), and if it is grammatical, to present the YACC-based struture for human or other automatic processing (Nora's glosser takes the parser output for example and does a simple word-for-word translation. It may eventually ave more sophisticated syntax-based tranlsation capabilties, but may never try to resolve "ri". lojbab