Re: [lojban] RECORD: place structure.

[pycyn@aol.com]

In a message dated 00-05-19 13:01:16 EDT, you write:

<< 
 On the other hand, in Lojban there are many more sumti tcita <BAI> than in
 Loglan, such as ga'a (zgana, as observed by) or mu'u (mupli, for example),
 a total of 64 of them according to my notes, in addition to the tenses
 <PU>.  They are etymologically and semantically closely related to
 particular gismu, and the effect of tagging a bridi with one of these
 prepositional phrases is pretty easy to predict regardless of the selbri
 (predicate) in the bridi. In addition, fi'o...[fe'u] allows any selbri to
 be used in a <BAI>-type phrase.
 
 This is in contrast to natural languages where the meaning of the
 predefined cases (genitive, dative, ablative, etc.) has just enough
 regularity to confuse the learner, who ought to be just memorizing their
 use with each predicate individually.
 
 From time to time in the past, the fashion has shifted between giving
 numbered places to "all" "essential" arguments of a gismu, versus letting
 the more outre' arguments be served by <BAI> phrases.  At the time of
 baselining the pendulum was stuck on the side that is not my favorite.  Of
 course, the speaker is not required to use numbered places if his judgement
 of style suggests that a <BAI> phrase would be better. >>

Well, some would prefer fewer places ("under condition" seems to turn up a 
lot at the end of lists, say), other would prefer more.  The point here is 
that the BAI list works so well just because it is based in the predicates 
and so has a thoroughly fixed meaning that can be learned without much extra 
effort, as opposed to the minimal-place predicates and prepositions or cases 
or whatever, where the meanings of the latter have already to be learned anew 
with each predicate.  The altter seems to be less efficient in even the short 
run