[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
At 12:00 PM 4/30/03 +0100, And Rosta wrote:
Someone someday is going the face up to the fact that de facto
grammar of {kau} is that it can follow any question word, but
not meaningfully anything else, and write a grammar accordingly,
using whatever formalism best gets the job done. And so forth, for
the entire de facto grammar of the language.
First of all, UI is a single selma'o in name only, just as PA is. Each really is several selma'o all with essentially identical grammars but often drastically different semantic natures. Thus we have the groupings UIn in the cmavo lists, though even they were not analytically determined, but rather were groupings for lexicon learning benefit to make studying members of a large group easier by breaking them down into semantically similar groups as well as a miscellaneous group. Both UI and PA originally had explicit grammars for restricting the ordering of these subgroupings, but the restrictions proved insupportable.
Next, kau (as designed) is not limited to question words, even though that may be what current usage norms are.
The reason kau was put in UI was that at the time it was added, we kept finding new places where it was useful, and could not come up with a reason why semantics might not be able to apply it to ANY usage.
In most cases, kau following non-question words is similar in meaning to kau following the question word corresponding to the selma'o that is marked. Early history will likely show "dakau" used as often as makau, and there may even have been usages of things like "mikau". Thus it is only when kau follows a word *or construct* that has no corresponding question-word that there is no clear basis for meaning. The fact that kau can apply leftwise to a construct as well as to an individual word made it quite UI-like (as well as quite free-modifier like, I'll admit).
Since we do not have a GRAMMATICAL restriction against any selma'o having a corresponding question word, there can be no grammatical restriction on kau. Thus it needed to go in UI rather than as a selma'o of its own in the free modifier grammar.
NAI was NOT the same, because NAI already had two roles in the language. Its use in preparsed logical connectives was deemed incompatible with its use in UI. "a nai" is neither a contrary nor contradictory negation of a.
I all likelihood, if NAI had not been used in logical connectives, it would indeed have been merged with CAI, which basically is the same as merging with UI. However, I note that in early Lojban formal grammar, UI grammar WAS explicitly in formal grammar, including the substructure of combinations of UI, CAI, and NAI. It was removed only when it was argued (possibly by Cowan, but this may have predated him) that properly we would have to put a rule for attitudinals after every construct, and the version of YACC that we had simply couldn't handle that many rules. (We were constantly pushing against the limits of available YACC versions throughout the 16-bit era).
People can say that the multiple use of NAI violated the design principles of the language, but the bottom line is that it was solidly enshrined in Loglan long before I got involved, and the sacredness of that particular design principle was less firmly enforced in JCB's world than in redesigned Lojban.
I'm not sure whether I can elucidate the argument clearly, but I see too much similarity between the argument (Jorges?) for nai in UI, with the arguments for and against prepositions on place structures. (JCB experimented with case tags for places in TLI Loglan and we rejected them as being the reintroduction of prepositions.)
While to some extent it is true that "nai" is polysemous between contrary, contradictory, and something else (in logical connectives), the proper solution in the past for Lojban has been to split the usage into multiple cmavo, NOT freeing up the grammar. The polysemy of nai is quite a bit short of the polysemy of prepositions in natural language; for one thing, I think that nai is consistent in meaning with any one selma'o (or group of words within a complex selma'o - Im not going to try to figure out whether kaunai is a scalar negation of kau just because uinai is a scalar negation of ui); defining that polysemy in a dictionary is thus a tractable problem, even though it is a relative pain.
lojbab -- lojbab lojbab@hidden.email Bob LeChevalier, President, The Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273 Artificial language Loglan/Lojban: http://www.lojban.org