Date: Sat, 14 Feb 1998 19:57:19 -0500 (EST) Message-Id: <199802150057.TAA28784@locke.ccil.org> Reply-To: Chris Bogart Sender: Lojban list From: Chris Bogart Subject: Re: perversity squared X-To: lojban To: John Cowan X-UIDL: 7a27caa3c31057e8a5247cb08a79ff32 X-Mozilla-Status: 8011 X-From-Space-Date: Tue Feb 17 10:17:14 1998 X-From-Space-Address: - I hadn't thought of the baseline/usage schedule of Lojban in quite the = terms Lojbab=20 is talking about before. The idea reminds me of an optimization = technique called simulated annealing. You have one way of optimizing a thing (logical = analysis and=20 argumentation) that works for a while but tends to get stuck in a dead = end;=20 that's when you kick in another method that shakes things up some = (giving users latitude to evolve the language in a more usable direction), then you go = back to the first method and apply it some more (logical reanalysis of what the = users did with the language) Going back and forth between the two can give you = better results, fater than what you would have acheived with either method = alone. >Thus I envision that people will memorize the gismu themselves, along = with >keywords and/or other synonyms that will give them a gestalt sense of = the >meaning. In many cases they will also remember the number of places or = at >least the roles of some of the places - not all, and not with semantic >precision. If people do things the way you suggest, I bet it will lead to = significant regularization of the gismu places. =20 >If people always or usually add a BAI tagged place to a gismu, that = place >should be added without requiring the BAI. =20 That might depend on why they use BAI. I have a tendency to use "bau", = for example, even with gismu that already have a language place, because = I have "bau la lojban" stuck in my head as a set phrase, and it's easier = to stick it on to a predication using "tavla" than to remember to use = "fu" or topsi-turvify the sentence to bring the x5 into a more = accessible place. Even with a BAI tag that's not redundant, it may be that it's better off = as a BAI tag than as a proper place, if there are already twenty-leven places and the new place woudl = be lost in a sea of zo'e's and bei's without the BAI. I'd say, if the gismu place structures are to be set by usage, then we = could assume that if something should be a place, it'll become a place, = and if something gets added a lot with BAI, it should stay that way. co'o mi'e kris