From thanatos@dim.com Tue Feb 26 22:34:51 2002 Return-Path: X-Sender: thanatos@dim.com X-Apparently-To: lojban@yahoogroups.com Received: (EGP: unknown); 27 Feb 2002 06:34:50 -0000 Received: (qmail 31318 invoked from network); 27 Feb 2002 06:34:50 -0000 Received: from unknown (216.115.97.171) by m5.grp.snv.yahoo.com with QMQP; 27 Feb 2002 06:34:50 -0000 Received: from unknown (HELO supernova.dimensional.com) (206.124.0.11) by mta3.grp.snv.yahoo.com with SMTP; 27 Feb 2002 06:34:45 -0000 Received: from p36.3c04.pm.dimcom.net (p36.3c04.pm.dimcom.net [206.124.3.196]) by supernova.dimensional.com (8.11.4/8.11.4) with SMTP id g1R6Yhi12871 for ; Tue, 26 Feb 2002 23:34:43 -0700 (MST) To: lojban@yahoogroups.com Subject: Re: [lojban] go'i: repeated referents or just sumti? Date: Tue, 26 Feb 2002 23:40:40 -0700 Message-ID: References: In-Reply-To: X-Mailer: Forte Agent 1.7/32.534 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable From: EWC X-Yahoo-Group-Post: member; u=45881577 X-Yahoo-Profile: thandim2000 X-Yahoo-Message-Num: 13414 On Tue, 26 Feb 2002 23:07:02, "Jorge Llambias" wrote: >la tanatos cusku di'e >>The gimste definiton of {sumti} says that lo sumti is text, so if you're >>speaking in Lojban about sumti you have to quote words. >So it does, you're right. The place structure does not make >much sense then, It wouldn't be the first place structure not to make complete sense. ;) >because the form "le mlatu" is a sumti form >irrespective of any selbri, whereas a cat is not an argument >unless and until there is a relationship that can accept cats >as arguments. It smells to me like a bootstrapping problem. A natural Lojban speaker might not think of things being arguments to relations at all. Things are what they are, and sumti, selbri, and bridi are just pieces of language used to express that. Cats are never arguments, they're cats and whatever else they are. {ta mlatu} just means that is a cat, not that fills the x1 place of mlatu. This is a hard point for me to make, so I doubt I'm making it clearly. Sumti, selbri, and bridi are pieces of language and not what those pieces of language are used to refer to. We refer to the things referred to by just using the language, not talking about pieces of the language. >>Similarly, lo >>bridi are also text. >Yes, the whole system seems to be pretty messy if we >take the gi'uste at its word. I think it may be consistent, and we just need to use it consistently.=20=20 If we want to ask "what is the relation between me and you", we ask {mi do mo}, not {ma selbri fi zo mi ce'o zo do} unless we're actually asking about a piece of text in a class on Lojban grammar. So I think having sumti, selbri, and bridi all be text works fine for speaking about Lojban texts in Lojban, as it should be; it's just not very convenient for talking about Lojban semantics in another language. >>That makes {sumti} a grammatical category like >>"noun". I'm not a noun, but "I" is. >> >>.i mi na sumti .ijeku'i zo mi go'i >i zo mi sumti ma? .i zo mi sumti lu mi du mi =20 which is entirely different from saying .i mi du mi >> .i le su'o mlatu cu catlu mi >> .i le su'o mlatu cu catlu le gerku >I agree about the logic, though the explicit {su'o} does not make >it any more palatable to me that they be different cats. If it were {su'o le mlatu} you wouldn't expect the same at least one cat each time though, right? You're happy to reapply {su'o} as an outer quantifier but not as an inner? >>If you were a >>computer interpreting those statements you couldn't assume the "all of >>at least one described as cat"s were the same. >I agree, but a well programmed computer should strongly infer >that they are, in my opinion. I'd only go so far as to say that it should ask if they are the same, or possibly allow that they are the same as long as that doesn't contradict anything else. It shouldn't draw any conclusions that rely on their being the same, however. >>I'll still happily accept that it's convention that the speaker >>shouldn't change mental "described-as" groups, changing {le mlatu} to >>meaning "all of all the things I'm described as cats in this >>discussion", but the underlying logic doesn't require it. > >"all of all" does not add information, on the contrary, it loses >information, at least if you agree with me that "all" does not >have existential import. Oops. I left out an "at least one" there. It's ugly in English, but how about, "of all of the at least one things I describe as a cat in this discussion, all ..."? {su'o le mlatu} would then be "of all of the at least one things I describe as a cat in this discussion, at least one ...". That's the interpretation that keeps all the cats in mind the same for the entire discussion, and how we actually want to use {le mlatu}, as if it selected from all things described as cats in this discussion in the same way {lo mlatu} selects from all things actually cats in existence. >>Pro-sumti are >>there to force repetition of referents if needed, after all. > >Yes, though they have their own complications too. That brings up a question I don't remember being addressed. Does pro-sumti assignment happen before or after the outer quantification? I'd guess that ko'a in {re le mlatu goi ko'a} would refer to all the cats in mind, while in {re le mlatu ku goi ko'a} ko'a would refer to the two cats out of all in mind, the relative clause being either inside or outside the description and therefore either before or after the outer quantification. Hmm, but then what about BY pro-sumti used without assignment? {re le mlatu cu blabi. i pa my. sipna}. Is that one of the two white cats or one of all the cats in mind in the first statement? --=20 EWC