From pycyn@aol.com Tue Sep 25 09:31:23 2001 Return-Path: X-Sender: Pycyn@aol.com X-Apparently-To: lojban@yahoogroups.com Received: (EGP: mail-7_3_2_2); 25 Sep 2001 16:31:23 -0000 Received: (qmail 61142 invoked from network); 25 Sep 2001 16:22:26 -0000 Received: from unknown (10.1.10.26) by l8.egroups.com with QMQP; 25 Sep 2001 16:22:26 -0000 Received: from unknown (HELO imo-m01.mx.aol.com) (64.12.136.4) by mta1 with SMTP; 25 Sep 2001 16:22:26 -0000 Received: from Pycyn@aol.com by imo-m01.mx.aol.com (mail_out_v31_r1.7.) id r.66.14cdf94e (17085) for ; Tue, 25 Sep 2001 12:22:10 -0400 (EDT) Message-ID: <66.14cdf94e.28e20931@aol.com> Date: Tue, 25 Sep 2001 12:22:09 EDT Subject: Re: [lojban] Set of answers encore To: lojban@yahoogroups.com MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="part1_66.14cdf94e.28e20931_boundary" X-Mailer: AOL 6.0 for Windows US sub 10535 From: pycyn@aol.com --part1_66.14cdf94e.28e20931_boundary Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: quoted-printable In a message dated 9/25/2001 10:34:50 AM Central Daylight Time,=20 arosta@uclan.ac.uk writes: > #But then next morning, looking at it again, I see that it is very=20 > different,=20 > #for I doubt that John (even this one) even thinks about "the extension o= f=20 > "=20 > #some property. In fact, I doubt that most people, who use indirect=20 > question=20 > #all the time, would even understand the locution. So, if the property i= s=20 > #within the scope of the believing, where, because of intensioonality, i= t=20 > has=20 > #to be that property and not something incidentally equivalent to it, the= n=20 > I=20 > #would say that it was very rarely the case that anyone had an opinion=20 > about=20 > #who the first American President was. But, of course, the other version= ,=20 > #which moves the property outside still works ok. >=20 > Jorge raised this objection at the time that I originally made the propos= al. > My answer is that if the extension-claim analysis correctly characterizes > the logic of indirect questions, then if John knows that 'Bill' is the=20 > answer > to 'Who went', then John knows that {Bill} is the extension of the > category of goers. >=20 Well, it is quite possible that John knows that Bill went, without knowing= =20 that "Bill" is the answer to the question "Who went?" for the very reason y= ou=20 note later, that he never thought of the question. It is also not at all=20 clear that the equation of the two things John knows works intensionally. The first problem does offer some evidence for the set-of-answers theory,=20 since even if John never thinks of the question "Who went?" he does know th= e=20 answer to that question, since "Bill went" is just that answer. I suspect= =20 that this fact can be mechanically transformed into an extension-claim=20 version, though doing so makes the analysis more wordy apparently. (The=20 second problem does not arise for set-of-answers.) This seem complex compares to {la djan djuno lo du'u la bil patfu makau},=20 which says he knows some, but not necessarily all, the answers. That it ma= y=20 turn out to be all is an open case. I suppose the simplicity is in the lac= k=20 of unpacking, but set-of-answers doen't need much unpacking, since it stays= =20 at about the same level, without metalanguage: da poi cmima lo'i du'u la bi= l=20 patfu makau zo'u la djan djuno da Notice that there appears to be no de-intensionalization problem here. Your version seems to say merely that John knows "Bill is the father of ...= "=20 has an extension, which seems a) not likely to be somehting John thought of= ,=20 and, b) if he did -- or if you want it not to matter that he did, is too=20 trivial to help explain what "John knows who Bill is the father of" means. = I=20 think that something ahs to be done with that floating {de} to make any sen= se=20 at all and I don't feel comfortable enough with the extesnion-claim format = to=20 suggest what that something might be. --part1_66.14cdf94e.28e20931_boundary Content-Type: text/html; charset="ISO-8859-1" Content-Transfer-Encoding: quoted-printable In a message dated 9/25/2001 10:34:50 AM Central Daylight Time, arosta@uc= lan.ac.uk writes:


#But then next morning, l= ooking at it again, I see that it is very different,=20
#for I doubt that John (even this one) even thinks about "the extension= of "=20
#some property. In fact, I doubt that most people, who use indirect que= stion=20
#all the time, would even understand the locution.  So, if the pro= perty is=20
#within the scope of the believing,  where, because of intensioona= lity, it has=20
#to be that property and not something incidentally equivalent to it, t= hen I=20
#would say that it was very rarely the case that anyone had an opinion = about=20
#who the first American President was.  But, of course, the other = version,=20
#which moves the property outside still works ok.

Jorge raised this objection at the time that I originally made the prop= osal.
My answer is that if the extension-claim analysis correctly characteriz= es
the logic of indirect questions, then if John knows that 'Bill' is the = answer
to 'Who went', then John knows that {Bill} is the extension of the
category of goers.


Well, it is quite possible that John knows that Bill went, without know= ing that "Bill" is the answer to the question "Who went?" for the very reas= on you note later, that he never thought of the question.  It is also = not at all clear that the equation of the two things John knows works inten= sionally.
The first problem does offer some evidence for the set-of-answers theor= y, since even if John never thinks of the question "Who went?" he does know= the answer to that question, since "Bill went" is just that answer.  = I suspect that this fact can be mechanically transformed into an extension-= claim version, though doing so makes the analysis more wordy apparently. &n= bsp;(The second problem does not arise for set-of-answers.)

<However, there is a valid case not quite covered by my extension-cl= aim
analysis. An example is where John knows that Chelsea is Bill's
daughter but doesn't know that Bill has no other daughters [by Hillary,
that is, I hasten to add, having watched the splendid Primary Colours
twice in the last week). So, as it were, John knows who Bill's daughter= s=20
are, but doesn't know he knows. The extension-claim analysis can=20
handle John's actual beliefs thus:

=A0=A0 la djan djuno tu'odu'u da cmima tu'o -extension be tu'odu'u la b= il patfu ce'u>

This seem complex compares to {la djan djuno lo du'u la bil patfu makau= }, which says he knows some, but not necessarily all, the answers.  Th= at it may turn out to be all is an open case.  I suppose the simplicit= y is in the lack of unpacking, but set-of-answers doen't need much unpackin= g, since it stays at about the same level, without metalanguage: da poi cmi= ma lo'i du'u la bil patfu makau zo'u la djan djuno da
Notice that there appears to be no de-intensionalization problem here.

<And the version where we deintensionalize our description of John's
knowledge can, I very very very tentatively think, be done thus:

=A0 da poi ke'a du'u de -extension tu'odu'u la bil patfu ce'u zo'u la d= jan djuno da

I am very much not convinced that this solution is valid, but if it isn= 't, it's just a
further instance of the more general problem of how to refer to se djun= o
and se jinvi extensionally rather than intensionally, and any solution = of
the more general problem will also resolve the residual problems with t= he
extension-claim analysis.>

Your version seems to say merely that John knows "Bill is the father of= ..." has an extension, which seems a) not likely to be somehting John thou= ght of, and, b) if he did -- or if you want it not to matter that he did, i= s too trivial to help explain what "John knows who Bill is the father of" m= eans.  I think that something ahs to be done with that floating {de} t= o make any sense at all and I don't feel comfortable enough with the extesn= ion-claim format to suggest what that something might be.
--part1_66.14cdf94e.28e20931_boundary--