Return-Path: From: cbmvax!uunet!pic.ucla.edu!jimc Return-Path: Message-Id: <9105221903.AA12264@whistler.pic.ucla.edu> To: lojban-list@snark.thyrsus.com Subject: What binxo means Date: Wed, 22 May 91 12:03:43 -0700 Status: RO X-From-Space-Date: Wed May 22 15:52:27 1991 X-From-Space-Address: cbmvax!uunet!pic.ucla.edu!jimc The original discussion was on the need for the reciprocal operator soi, but now it's on the adequacy of binxo "become". Let's translate "my rat died" using the interpretation of binxo that John Cowan gave, with a backmap of what this really means in its full glory: lemi ratcu cu binxo lo morsi At present my (present) rat is identical to at least one small stiff, but formerly there was no stiff that my (former) rat was identical to. (The "at least one" part comes from binxo, not lo). Now suppose someone came up with this sentence from the lesson 1 exercises: la banthas. du lo mlatu Bantha is (sic) a cat We would tell the student "that's very nice, but the preferred way to say that is..." la banthas. mlatu Bantha cats In other words, Lojban has predicates in its foundation, and prefers to use them. Now I say the same thing about binxo; the meaning "become" *ought* to be rendered with a predicate analogously to "la banthas. mlatu". However, I see no gismu with a more suitable set of arguments than binxo. In a previous message I said "binxo means 'x1 changes so event x2 becomes true'", and John didn't like either the content or the event of saying. Let me half apologize and half not apologize. I don't mean to speak authoritatively so as to imply "the LLG has adopted (the above) meaning for binxo", which is manifestly false. However, as has been said many times, the focus of LLG work has been mainly on morphology and grammar, and the semantics of the words, specifically the place structures, are to be worked on in a near-future phase. For the present the Old Loglan definitions are used, with a few hacks on the worst infelicities. Thus when one sees binxo bix bi'o become become...under conditions... one has very little guidance for what fits in the "..." Into this vacuum comes I, having sweated over exactly this problem since 1979 (gee, is it that long?) It is hard not to say "binxo means..." where politeness requires "binxo ought to mean... convince convince convince", particularly when the focus is not on the meaning of binxo but on how a reciprocal relation suddenly vanishes when a second predicate gets into the act. In the future I will try to say "ought to mean", but I plan not to clog up net bandwidth with convincing arguments unless someone asks me specifically, so as not to detract from the focus of the discussion. Now, back to the content. When I define "x1 changes so event x2 becomes true" it's to be interpreted very much like my rendition above of sumti binxo: lemi ratcu cu mrobi'o My rat died lemi ratcu cu binxo (le nu lemi ratcu cu morsi) lemi ratcu cu cenba My rat changed .ija lemi pu ratcu pu na morsi My former rat wasn't dead .ija lemi ca ratcu ca morsi My present rat is dead In other words, for each referent coming into x1 two predicate relations (or sumti identity relations in John's place structure) are synthesized, one for an earlier process step and one later. (The process is usually timelike but not always.) This is how an event "changes to become true"; there are two events, one is false and one is true. You can see how attractive "mrobi'o" is to express the idea of "died", but either the user has to memorize the meanings and places of a zillion individually crafted lujvo (one for each non-become-meaning gismu and lujvo, plus similar lists for each of the gismu common in compounds, like "stop", "force", "believe", etc. etc.), or there must be rules that let the user boil down a lujvo (or tanru) to a collection of phrases like "x1 binxo le nu x1 morsi". And in order for the synthesized phrases to mean something, it is necessary to replicate arguments into them -- a capability which turns out to be useful for explicit event arguments too. One item about binxo that isn't too clear is what gets replicated, words or referents. In the examples above I replicated the words of x1, once with "pa" and once with "cu" to get "my former/present rat". Given my emphasis on postponing the calculation of referent sets to the very end, this is the choice I prefer. An attractive alternative is to consider each x1 referent to have a history or process trajectory, and in the earlier-later relations to trace along that history. While conceptually attractive to humans, this model would be quite difficult to program in a database handler and would require specialized support just for binxo. However, when the words are replicated and the earlier-later copies are individually bound they will latch onto the history automatically, if there is one, and the individual support would be confined to generating the two relations, a much easier task. -- jimc