Lojban, as usual, str= ives for the most of all possible worlds. Thus, for arguments to pred= icates, it has a place system -- several, in fact -- and a case system of s= taggering complexity. The problem -- if it is one -- with the place s= ystem, is that the places are not uniform. To be sure, the first plac= e is generally the subject, as that would be informally understood, and the= second is often the direct object (ditto). From there on (and even t= here occasionally) the places wander every which way, although there are so= me families of concepts which give rise to predicates all of which share mo= st of the same places in the same order -- with shocking exceptions and occ= asional late additions. Then there are at least two ways to change th= e surface order of the arguments while keeping the original place information. In addition, there are the prepositions (mainly, but fa= r from exclusively BAI) that directly convey the information buried in the = places of particular predicates and which can be used instead of or in addi= tion to many positional roles. I am not sure that all subject and dir= ect object notions are covered by these critters, but I know that the most = common are -- and are frequently used. Lojban does not have -- and ne= ver will, pretty obviously -- affix cases, but the reputed psychological di= fferences between prepositions and affixes is not significant here (I think= ). It is worth noting, however, that most predicates co-occur with on= ly a limited number of prepositions, usually some involved already in their= specifications. There are also an array of prepositions that seem to= occur with just about any predicate, and may be deserving of special atten= tion. As for an optimal system (not one I would propose at this point for Lojban), each predicate would have one or two defined place= s and a list of likely special prepositions (the later places or ones thoug= ht of for those places). The defined places would be the expected one= s, subject and object in their predicate-relative senses, and everything el= se would be added explicitly. Several people seem to write this way a= lrady.

= From: la gleki <gleki.is.my= .name@gmail.com>
To: lojban@googlegroups.com
Sent:<= /span> Wednesday, March 27, 2013 12:17 PM
Subject: Re: [lojban] A Discussion of the Lojban System = of Place Structure

On Wednesday, March 27, 2013 8:52:52 PM UT= C+4, tsani wrote:
= Something that bothers me about your paper, which as it stands is quite goo= d, is that you only deal with an extremely small portion of Lojban. You don= 't touch at all on lujvo, which are what place structures are really for. I= n the case of purely compositional lujvo, e.g. kansi'u, the place structure= lets you infer without any ambiguity or metaphor what the resulting place = structure of the lujvo will be, based solely on the knowledge of the compon= ent gismu. As for memorizing the place structures of component gismu, one m= ust simply realize that Lojban already *has* a case system. Simply put, rat= her than having named cases, we have numbered cases, and in many&nbs= p;cases, there is a lot of parallelism between gismu of the same semantic category.

Consider {morji}, {djuno}, {smadi}, {sruma}, {jijnu}, {= jimpe}, {senpi}, {birti}, {krici}, and {jinvi}.
morji =3D x1 reme= mbers fact x2 about x3
djuno =3D x1 knows fact x2 about x3 by epi= stemology x4

smadi =3D x1 guesses x2 about x3
jijnu =3D x1 intuits x2 abo= ut x3
jimpe =3D x1 understands fact x2 about x3
senpi = =3D x1 doubts that x2 is true
birti =3D x1 is sure that x2 is tru= e

krici =3D x1 believes that x2 is true about x3
jinvi =3D x1 = opines that x2 is true about x3 on grounds x4

In t= he "knowledge" family, which above isn't even complete, one immediately not= ices the parallelism. This family along with the "family" family (family-re= lation =3D x1 is the <type of relationship> of x2 by bond/tie x3; e.g= . bersa) and the foods and animals family (food/animal =3D x1 is a <food= /animal> of type/species/etc. x2) make up an extremely large portion of = the gismu list.

Now, I concede that even within a family, there are som= e exceptions, such as birti and senpi, which don't have an "about" place at= all, or jinvi and djuno which are the only ones with an x4. Even in the fo= od and animals family, there are some notable oddballs such as lanme ( =3D = x1 is a sheep of species x2 of flock x3). However there being a difference = in the number of places *usually* isn't that much of a big deal. If someone= accidentally fills in an x3 of birti, any decent listener will understand = what is meant, due to the overarching systematic nature of the family to wh= ich it belongs. On the other hand, when lujvo are made (according to the ru= les, that is) oddball places may unfortunately need to be included, which c= an lead to unexpected place structures in the resulting lujvo.

e.g. jboxlajivdunsi'u =3D x1 are equal in that they opi= ne that lojban is bad for x2 by standard x3 on grounds of belief x4.
<= div>
If one forgets that xlali has a standard place in the x3= , then if they try to fill in the belief place of jinvi, things can get mes= sy.

Still, I don't know how your proposed case system would= cooperate with lujvo (or would it simply drop lujvo from the language?) So= I can't say that the status quo is any better or worse.

Finally, semantic families don't always work. {cpedu}, {mind= e}, and {picki} are in the same family, I'd say, but have place structures = divergent in rather irritating ways.

cpedu =3D x1 requests x2 of x3 in manner x4

minde =3D x1 orders x2 to x3
picki =3D x1 begs x2 to x3

However, if one takes into account the emphasis differ= ence, (when ordering something to happen, the person whom is ordered is mor= e important, and therefore moves closer to the front,) then it becomes easi= er to remember. Also, cpedu2 being the action requested is more useful for = translating "He asked for a glass of water." Indeed, in that case, the pers= on of whom that action (giving a glass of water) is far less important (it = could be anyone, so in truth, we're just dropping the place entirely by con= text).

That being said, I strongly push towards efforts to reg= ularize the gismu list in order to make the numeric case system more self-c= onsistent. One of Lojban's major issues is a lack of self-consistency, as p= ointed out by gleki in his thread about "four different vocabularies."

I started that thread with a completely d= ifferent complaint, namely ....4 different but parallel sets of words/morph= emes for memorising.

However, the existence of this multiple vocabula= ries is not a problem intrinsic to the nature of Lojban, and is thus a whol= e other can of worms when contrasted with the issues with place structure.<= /div>

In sum, I believe that trashing place structure is over= all a bad idea.

Not at all. It *is* p= ossible to create a language where every gismu will have not more than two = places. May be Lojban being spoken by more people will try to change into s= uch thing. Whether such dialect of Lojban will be more easy to learn or not= is another question.

Place structure is central to Loj= ban's ideology, in my opinion, and removing it from Lojban would be like ta= king the Lojban out of Lojban.

mi zma= du do lo ni barda (I exceed you in property of being big) =3D> mi poi zi= lkarbi do cu barda (I compared to you is big).
This is how the th= ird place might be removed by those who hate it.

Named case systems are used in other conlangs, an= d if those are more pleasant to you, then perhaps those you should try thos= e out, too.

Even not touching the pre= vious example nobody can remove metaphors from our speech. If Daniel wants = to try those in Lojban why stop 'em?

<= br>
Still, = Lojban remains extremely interesting. It was not made to be easy, and compl= aints that its cornerstones are complicated are of little weight. As you le= arn more, you will see that there is a lot of regularity, and that this lan= guage still remains immensely more simple than a multitude of others.

.i mi'e la tsani mu'o

--
You received this message because you are subscribed to the Google Groups "= lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an e= mail to lojban+unsubscribe@googlegroups.com.
To post to this group, send email to lojban@googlegroups.com.
Visit this group at http://groups.google.com/group/lojban= ?hl=3Den.
For more options, visit https://groups.google.com/groups/opt_o= ut.

--
You received this message because you are subscribed to the Google Groups &= quot;lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an e= mail to lojban+unsubscribe@googlegroups.com.
To post to this group, send email to lojban@googlegroups.com.
Visit this group at http://groups.google.com/group/lojban?hl=3Den.
For more options, visit https://groups.google.com/groups/opt_out.

---420974808-1343082370-1364406983=:40791--