[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Ongoing discussion with TLI rep on Loglan/Lojban and logic - 2 of 3
Part 2 of 3+ of the exchange between lojbab and Randall Holmes
|Message 1:
|Date: Tue, 21 Jun 94 09:20:48 -0600
|From: Randall Holmes <holmes@diamond.idbsu.edu>
|To: holmes@diamond.idbsu.edu, lojbab@access.digex.net
|Subject: Re: Lojban
|Cc: nsn@krang.vis.mu.oz.au
|
|What is the address of your ftp site?
|
|I can tell something about the kinds of refinements you folks have been
|working on (at least, what you were working on a while back when I got
|the material you sent me). Mostly, they are good. I have convinced TLI
|and hope eventually to convince you folks that ME should have a more
|precise meaning!
|
|I'm sorry to hear about the position of TLI on translation between the
|languages; the automatic translation software I suggested would probably
|fall in the same category (unless TLI owned it, I suppose). I have
|nothing to do with that side of things; I have gently made it known that
|I think we have no secrets worth protecting and that everything in the
|grammar ought to be available to every speaker of the language (but, in
|fact, this is true; a little experimentation with LIP will ferret out
|most secrets). But that is the most I can do.
|
|Also, is your software free, or do you charge for it?
|
| --Randall
|
|Mail>r
|To: holmes@diamond.idbsu.edu
|Subject: Re: Lojban
|
|I'd like to hear more on preciseness of meaning of ME. We have a defined
|place structure for the resulting predicate, if that helps, but the
|semantics is more-or-less vague as to what the actual claim is beyond
|that place structure (which I believe is
| x1 pertains to (ME argument) in property/aspect x2 (generally a LEPU
| abstraction)
|
|I am not sure that TLI's policy would forbid auto-translation. The
|argument may or may not be that our publishing a list of TLI words is a
|copyright violation. Using an encoded list of such words as data in a
|program seems one step less offensive.
|
|But note that although Parsons objected, and I asked for a response and
|told him the list was available on our ftp site, TLI has never asked me
|to remove it from the ftp site - the argument may simply have to do with
|putting it in a 'dictionary'. You can get the list from the ftp site,
|though. The file is (/pub/lojban)/draft/oldlog.txt
|
|The ftp site is ftp.cs.yale.edu
|
|The reference grammar papers, by the way are in /draft/refgrammar, not
|simply /refgrammar as I said previously.
|
|Our software is more or less available as Shareware. If it is up on the
|server, you can take it for free and send us what you think it is worth
|(we have suggested amounts for some things). For stuff not up on the
|server, we have a pricelist - most stuff has been around $25, but we are
|flexible on this. For example, our LogFlash (cf. MacTeach) is
|instrumented such that if someone actually works the words up to Under
|Control, and sends us the log and control files, we let the person have
|the software free as well as acopy of the draft textbook for free (or
|refund the money if it has been paid for). We desperately want to
|conduct a SCIENTIFIC test of JCB's word recgnition scores and see if
|they are actually meaningful, and this instrumentation will give us some
|first data for that test. (JCB did an informal test back in the 50s,
|but never published data or methodology).
|
|lojbab
|Cc: lojbab nick
|
|
|Message 1:
|Date: Tue, 21 Jun 94 14:15:22 -0600
|From: Randall Holmes <holmes@diamond.idbsu.edu>
|To: holmes@diamond.idbsu.edu, lojbab@access.digex.net
|Subject: Re: Lojban
|Cc: nsn@krang.vis.mu.oz.au
|
|I have argued, and Bob McIvor supports me, that the meaning of ME
|<designation> ought to be precisely "is one of the objects currently
|referred to by <designation>" In fact, I assumed from the outset that
|this was what it meant, because such a construction (_literal_
|conversion of a designator to a predicate) is so obviously needed.
|Thus, "Ta memi" actually means "That is me" (not an inspiring example,
|to be sure!). "Ta meSai" means "the object currently referred to by the
|letter variable S" (NOT S-shaped!), and "Ta me le to mrenu" means "That
|is one (or more, I suppose!) of the three men we are talking about".
|The real use of this comes in more complex contexts. The vague original
|meaning is still the meaning when ME <designator> is used as a modifier:
|
|Ta memi bekti "That's me" in the idiomatic sense of L1
|
|Ta mela Ford, tcaro "That's a Ford car"
|
|A complex example is the definition of a set (real logic!):
|
|let ...ba... stand in for some kind of sentence about ba which is
|the defining condition of our set:
|
|lea meba jio ...ba...
|
|"the set of all ba such that ...ba...
|
|The point is that lea constructs sets from predicates; it is easy to
|designate an indefinite object with the property specified by
|...ba..., but it is not so easy to come up with a coextensive predicate,
|without using ME in a precise sense.
|
|By the way, one of my correspondents from the posting on the Internet
|(who took "the pirate version" much too seriously, alas) made comments
|which seemed to indicate that some children may have learned Lojban; he
|alluded to creolization changes in the language. Any info?
|
| --Randall Holmes
|
|
|Mail>r
|To: holmes@diamond.idbsu.edu
|Subject: Re: Lojban
|
|Other than Jenny, my kids are the only ones that have learned any
|Lojban, and they have not learned it well enough to speak much of
|anything, much less affect creolization changes. But it may happen yet.
|They are clearly learning it more as a native speaker would - by
|assimilation rather than direct teaching.
|
|If I understand your explanation, I have one problem with your
|definition of ME: I don;t like the idea that it should mean something
|different as a modifer than as a main predicate. The two meanings OUGHT
|to be consistent.
|
|I don't think that we have a problem defining sets such that we need to
|use ME, either, but I don't see enough in your example to clearly tell
|me what you mean by "indefinite object with the property specified by
|...ba..." vs. coextensive predicate.
|
|Lojban has two mechanisms that are used for describing sets. We have a
|descriptor, with grammar of "le", which is "lo", that refers not to the
|intensional described thing, but to themembers of the set of things that
|actually fit the predicate. This makes "lo" similar to "lea", but we
|use a different defaulty quantifier. "lea" as I recall it, would be "ro
|lo" where "ro" corresponds to TLI "ra"; i.e. "all of the memebers of
|the set that veridically fill the x1 of the predicate". You can use
|conversion to manipulate the predicate in question so that anything that
|can be described in a predicate (simple or complex) can be turned into
|such a veridical description.
|
|Corresponding to "lo" we have "lo'i" which is the set comprising the
|members "lo", and "loi" the mass of the members "lo" ("loi" is TLIs
|"lo"). We have a similar "lei" and "le'i" for masses and sets derived
|from intensional descriptions, and "lai" and "la'i" for massified and
|sets of named objects. (We also have two descriptors for typical and
|stereotypical: "lo'e" and "le'e".)
|
|All of these descriptors operate on predicates and turn them into set,
|mass, or individual arguments. We also have generalized converters that
|can work onm more complex sets: the members of LAhE include 3
|descriptors that can take a simple or complex argument and turn it into
|another argument of the any of the three types: set, mass, or
|individuals-of-set. Thus you can define a set argument by specifying
|its membership, then bracketing it with the appropriate LAhE, or you can
|massify that set with a different member, or specify one or more of the
|individuals of the set using the third member of LAhE.
|
|I think between these two structures, we have the capability of handling
|all the types you are descrining in your message.
|
|Now, I just thought of the other thing I don't like about your version
|of ME. Why does it refer to "one or more out of" the set. To me, "me le
|to mrenu" MUST be saying something about BOTH of the two men ("to" does
|mean "two" and not "three", if I recall), and not just one of the two.
|This would apply to both the vague and any specific meanings - I cannot
|see that "me le to mrenu gu(?) tcaro is referring to cars associated
|with one of the two men, but rather to cars associated with BOTH of the
|two men.
|
|My wife mentions in passin that she did at one time argue for the
|interpretation of ME that you favor, but was 'outvoted'. I think we
|have the mechanisms to cover whatever you can come up with as examples,
|but if not, or if we have to get real clumsy, you might win us over.
|
|(Of course Jon Cowan and Nick Nicholas, who are also reading this
|discussion, might chime in here with some opinions (hint, hint!).
|
|lojbab
|Cc: lojbab nick
|
|Message 5:
|Date: Tue, 21 Jun 94 23:08:59 -0600
|From: Randall Holmes <holmes@diamond.idbsu.edu>
|To: holmes@diamond.idbsu.edu, lojbab@access.digex.net
|Subject: Re: Lojban
|Cc: nsn@krang.vis.mu.oz.au
|
|The fact that ME {designation} has the meaning I describe in modifier
|position is not a separate fact about it; it is a consequence of the
|"vague" meaning of the modifier construction itself.
|
|I was being misleading when I said that "me le to mrenu" means "is one
|or more of the three men"; I was thinking about the effect of a plural
|argument on it. It is simply a predicate which applies to exactly those
|three men that I mean.
|
|LEA preda means "the set of everything such that <preda>" in Loglan; the
|quantifier is ra.
|
|I have found repeatedly that it is useful to be able to construct a
|predicate which covers the objects to which a designation applies; the
|examples I gave were not exhaustive, nor do I claim that they could not
|be reproduced in other ways. Why did your wife favor the meaning I
|promote?
|
| --Randall Holmes
|
|Mail>r
|To: holmes@diamond.idbsu.edu
|Subject: Re: Lojban
|
|RH> I was being misleading when I said that "me le to mrenu" means "is one
|RH> or more of the three men"; I was thinking about the effect of a plural
|RH> argument on it. It is simply a predicate which applies to exactly
|RH> those three men that I mean.
|
|I'm not sure that your statement here is clear. If the predicate
|applies to "exactly the three men", then it could mean:
|
|a) the set of the three men would go in the x1 place (which is the only
|place of the predicate); a partial set would not be correct
|
|b) the mass of the three ment would go in the x1 place;
|
|c) the three men as individuals could go in the x1 place, but it would
|have to be all three men in order to be a true claim
|
|d) one or more of the three men could go into the x1 place - it would
|not have to be a compolete specification of the individuals.
|
|I think I agree as to the meaning of LEA - that would be LOjban "lo'i".
|As originally proposed in TL, it was unclear whether it was a set
|descriptor or a veridical set-of-individuals descriptor with default
|quantifier "ra". The difference is determined by the types of claims
|that can be made about "lea mrenu", for example. If "lea mrenu" has set
|properties, and the types of predicates in can be used in are things
|that refer to such properties, than it is a true "set" argument. But
|early usage was more along the lines of "lea mrenu" are characterized by
|penises, which is not a property that a >SET< would have.
|
|The original definition of "me" was specifically as an inverter for a
|description, and I think that Nora therefore presumed the definition
|would be that of a converter - having no semantics but merely
|mechanically changing an argument into a 1-place predicate which that
|argument was the complete value which would fill the place. The trouble
|was that neither JCB nor anyone else ever really used the word in this
|way, and its usefulness was supect unless it had the vague meaning.
|Meanwhile we HAD a use for the 'vague' version of "me" and could pin it
|down to a specific place structure that matched the bulk of historical
|usage.
|
|It would be trivial to add in a new member of "me" that had the formal
|conversion effect that you (and Nora0 favored, if it proved useful. I
|think "me" is more useful as it is, and I think the two uses (vague and
|formal) are sufficiently different that I would want to use two
|different words; we would use one of the disyllable CVVs for the less
|useful formal definition. I will see what John, Nick, pc, and Nora say
|about this, since it could be added now, and this is probably the last
|chance before our dictionary gets published.
|
|BTW, I got the new Lognet today. I think I disagree with JCB on why it
|is people have trouble expressing in Loglan/Lojban. It has nothing to
|do with logic and formal structures. It does have to do with the
|vagueness of English with regard to such formal strucxtures, and the
|fact that people who have trouble expressing these things are generally
|phrasing what they want to say in English and then tryong to translate
|the English words into Loglan, but they don;t know what the English
|words mean. If they express their ideas as predicates in the first
|place, there seldom is need for formal structures and or their
|manipulations,a t least in our experience. For example, in the "visible
|Loglan" text, one sentence appears to be a trabnslation of "in the room
|there was a blackboard with the words of my button written on it" or
|something like that. But this passive voiced existential English
|sentence just isn't what someone phrasing the sentence directly in
|Loglan would say. Rather they would say something like someone
|(ba/Lojban da) wrote the words (restrictive relative: printed on my
|button) on the blackboard in the room. Or omit the existential variable
|(which is irrelevant) by conversion or simple ellipsis. You just don;t
|need translations of "there is x" that often.
|
|Far more often in Lojban usage, I think people translate slowly because
|they want to create compounds with predictable place structures and
|meanings which convey exactly the meaning they want, and not merely the
|corresponding English semantics. Thus people really sweat over
|metaphors and we have put a lot of effort into trying to predict place
|structuires of the resulting compounds. Otherwise, the major difficulty
|has been simply that most 'adult English' expressions when turned into
|Loglan/Lojban, end up as compound, complex sentences with lots of
|relative clauses, abstractions, etc., and, while one can parse such
|sentences fairly easily if one stares at them for a while, you tend to
|stop and take a second look simply because you want to make sure all the
|terminators are there, etc.
|
|In short, the difficulty in Loglan/Lojban is that there is a mindset
|among those who try to express in the language to try to express with a
|precision that they would not try for in English.
|
|lojbab
|Cc: nick lojbab
|