Relevant t= o another thread, I notice that in posing and attempting to solve these sor= ts of problems, y'all regularly have implicit appeal to transformational gr= ammars of one sort or another.

= From: Jacob Errington <= ;nictytan@gmail.com>
To: lojban@googlegroups.com
Se= nt: Monday, July 30, 2012 7:26 AM
Subject: [lojban] Afterthought TAG Termset Connectiv= es

Do these exist?

The forethought version is gra= mmatical:

{.i nu'i ca gi zo .u'i ny nu'u gi zo .i'= i cy selsku}

Considering that afterthought TAG con= nectives take the form of <[connective] TAG bo> where [connective] is= the connective of the appropriate connective selma'o, e.g. {mi .e ca bo do= }, I would presume that the afterthought termset connective form would be &= lt;[sumti] pe'e je TAG bo [sumti]> where [sumti] is >=3D1 sumti. Unfo= rtunately, of course, this does not parse in jbofi'e.

Additionally, it would appear as though TAG connectives= such as {.e ca bo} are not valid utterances by themselves, whereas bare co= nnectives such as {.e} are. I'm not sure that this is desirable, but it doe= s make it impossible to answer {ji}-questions and such with tag connectives= . I can think of an example usage inspired by the CLL example of {ji}: = ;

{.i do djica tu'a lo tcati ji lo ckafi} "Would you like tea or coffee?= "

{.e ba bo} (English sucks for the answer:) "The former, then th= e latter."

(This example is incidentally unaffecte= d by the connective problem that usually arises when using connectives in n= on-top-level contexts like {tu'a ko'a .e ko'a}. The problem itself is that = it's arguable as to whether {tu'a ko'a .e ko'e broda} is equivalent to {tu'= a ko'a broda .ije tu'a ko'e broda} -> {lo su'u ko'a co'e cu broda .ije l= o su'u ko'e co'e cu broda} or rather {lo su'u ko'a .e ko'e co'e cu broda}.&= nbsp;

A similar, more drastic case occurs in {lo pendo be mi = .e do}, which we'd naturally assume is "the friend of both me and you" but = it's arguable as to whether {lo pendo be mi .e do cu broda} is equivalent t= o {lo pendo be mi cu broda .ije lo pendo be do cu broda} or {zo'e noi pendo= mi .e do cu broda}. My rationale behind supporting the latter interpretati= on is that using one descriptor should only create one description, whereas= the former interpretation creates two descriptions behind the scenes, name= ly {lo pendo be mi} and {lo pendo be do} which could be referring to entire= ly different things. Additionally although less importantly, the latte= r interpretation seems far more natural, not to mention that to achieve the= meaning of the former interpretation using the latter, one must necessaril= y employ the {zo'e noi} construction, which is unnecessarily cumbersome, wh= ereas with the latter interpretation, one must simply repeat {lo pendo} such that {lo pendo be mi ku .e lo pendo be do}. The only apparent = downside to the latter interpretation is that when using afterthought, in o= rder to repeat the main description, a lot of terminators might be required= , not to mention that if the description is fairly complex (a four-part tan= ru) it can become very annoying to repeat it.)

Returning to the main point of this post, considering t= hat termsets is a part of the language that everybody loves to hate, it's u= nlikely that afterthought tag termset connectives will ever see the light o= f day, but I don't think that the cmavo required for termsets are going to = disappear any time soon (at least not before mekso, for which there remains= proponents.) Therefore, I see no reason why this discrepancy between foret= hought and afterthought should exist. (Not to mention the impossible non-lo= gical bridi-tail afterthought connective, which I wanted to use just today.= ..) Could it be that afterthought tag termset connection was impossible due= to a parser limitation? What with the PEG being capable of all sorts of th= ings that the older grammars weren't, could this be implemented in the curr= ent grammar? As far as I can tell, making {pe'e JA TAG bo} grammatical brea= ks no usage at all.

mu'o mi'e la tsani

--
You received this message because you are subscribed to the Google Groups "= lojban" group.
To post to this group, send email to lojban@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegrou= ps.com.
For more options, visit this group at http://groups.google.com/group/lojban= ?hl=3Den.