[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >

[Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [jboske] Digest Number 217

Yeah. Back to answering digests.
On Tuesday, Aug 5, 2003, at 00:41 Australia/Melbourne, jboske@yahoogroups.com wrote: 


Message: 4
   Date: Sun, 3 Aug 2003 15:55:32 -0400 (EDT)
   From: Invent Yourself <xod@hidden.email>
Subject: Re: Re: Gadri for you

On Mon, 4 Aug 2003, Nick Nicholas wrote:

Bravo! A tour de force which gives us hope that we really do have the
intellectual firepower (specifically, le ka ce'u galfi ju jai gau stika) 
on board to bring this language to a superior resolution!


Would that I felt as optimistic...
We've switched sides -- as is our wont -- and now I am promoting a scheme on the strength of its being less damaging to the corpus! Your scheme does involve some minimal reassignment. (lu'o refers to the confusion that is 
"mass" just as much as does loi.)
But unlike loi, lu'o has seen much less usage, so its reassignment is safer. At any rate, an ambiguous lu'o with specialised LAhE for stuff and collective is something I could live with. Of course there is an alternative for collective: lo za'u lo broda , which is grammatical. Perhaps we can use that instead of LAhE as an unambiguous collective signal.
A LAhE DOES seem like the best option for signalling de re/de dicto. Also, 
this is somewhat orthogonal to the rest of the gadri.


There is definitely the whiff of {la'e lu'e} about de dicto, so I agree.


However, in
disambiguating loi and lo'e, these are pure gadri which never before
needed any LAhE, and, these are cases of honest conflation and not the
lack of a parallel or orthogonal structure. Therefore I still advocate the 
fractionation of loi and lo'e, and will adjust my scheme to co-opt your
results.
*shrug*. We'll see. I think maintaining the present conflation of loi is indispensable for any proposal that will be acceptable to the masses, but that's my opinion. 


Message: 9
   Date: Sun, 3 Aug 2003 19:33:28 -0400
   From: John Cowan <cowan@hidden.email>
Subject: Re: Re: essentials of a gadri system

Nick Nicholas scripsit:


Accepted. For my part, I have put my contribution on wiki, at
http://www.lojban.org/wiki/index.php/gadri%20report%2C%20aug%202003


A few notes:

1) It's not really true that "There exists an X" implies that X is
namable.  The number of real numbers, still more the number of
real functions of a single variable, vastly outruns the number of
names.  This doesn't really affect your argument.
"Name" is just an illustration of... uh, what is it, quiddity? You know better than me. At any rate, I'm adding a proviso; to make quiddity (or haeccity, I no longer remember the distinction :-( ) intelligible, it's as good an illutstration as any. 


2) Grammatically, LAhE3 can't be a LAhE, since those (like any other
sumti) can be quantified.  We need a new selma'o and a new grammar
rule if you *really* want unquantifiables.
Disruption to grammar should be minimised if this will pass. We are not used to seeing quantifiers on LAhE, the way we are on sumti. I would be contented to have the quantifier on LAhE3 be semantically ill-defined, rather than invent a new selma'o to have it also be syntactically ill-defined. If you, however, are willing to lend support to a new selma'o that excludes quantification, and if this does not backfire as perceived tinkering but is generally acceptable, I certainly have no objection to a syntactically unquantifiable LAhE. In all other regards, I do want this to be like LAhE though: it makes sense to speak of LAhE3 ci lo mikce (Mr Three Doctors).
But I gotta ask: have outer quantifiers ever been an issue for li? For la'e? I know And wanted a Unique quantifier (currently tu'o) to seal up the quantification of such gadri, but I cannot see why the default for them would be anything but tu'o anyway. 


3) I suggest that the "I fear" example be changed to "Joey fears"
forthwith, because first-person is nothing but a snare in these
examples.  It's rational for me to say "Joey fears the OBM" even
though I deny the OBM's existence. A fortiori, the other examples should be 
changed as well.


Done.


Message: 10
   Date: Sun, 3 Aug 2003 17:43:11 -0700 (PDT)
   From: Jorge "Llambías" <jjllambias2000@hidden.email>
Subject: Re: Re: essentials of a gadri system


la nitcion cusku di'e
Jorge, John and And have reproached you more civilly on lo'e than I can hope to. I don't care that you're used to using lo'e as a generic rather than how it has been officially defined by the LLG for years; if anything, I take a particular delight in combating such humpty dumptying. My recollection is that when I showed that the Generic needs to be orthogonal to ontological type, you went through all sorts of gematriya to hang on to just lo'e as an all purpose generic. I'll have no truck with it.
That there's usage of generic lo'e doesn't sway me either: as we've been saying all along, gadri have been so ill-understood and anglicised by Lojbanists that usage counts for little, and what can be made sense of in extant prescription (which mandates the collective/stuff conflation, before anyone brings that up) counts for all the more. All that past usage of Generic lo'e tells me is that there is a need for a Generic gadri vel sim., and lo'e looked available. That does not annul CLL-lo'e, because it does not prove CLL-lo'e is broken. You can keep hoping that the criteria for fixing are broadened beyond brokenness to elegance, and I will keep saying that you will never be granted that with this community of Lojbanists. You will never be granted it by me. We're dealing with reality here. 


lo'e makes
claims that are generic and intrinsic to the referent, i.e. "Lions live in Africa"; but claims that are not an intrinsic property of the referent 
should be made with "Mr" instead. "Lions live in Africa" is a claim in
some way definitional of lions, or at least characteristic of them. "I
like lions", or "I study lions", is not. Even though the referent is in a sense the same --- Lion-kind --- I think it dangerous to conflate the two kinds of claim. The syntax of the predications involved --- x1 vs x2 
position of lo'e --- might seem a way out of this; but of course that
is untenable in Lojban: lo'e cifno cu citka lo'e mirlrantelope is a
claim characteristic of both lion and antelope.


How is that property intrinsic? If lions became extinct, then antelopes
would no longer be eaten by lions. Would that mean that they are no
longer antelopes?
Intrinsic was misspoken for characteristic. I was thinking of the fact that "I study lions" cannot be a property characteristic of lions, and definitional of a prototype model of lions. 


Message: 11
   Date: Sun, 3 Aug 2003 22:35:05 -0400 (EDT)
   From: Invent Yourself <xod@hidden.email>
Subject: lo and intension (was: essentials of a gadri system)

On Mon, 4 Aug 2003, Nick Nicholas wrote:




Message: 1
   Date: Sat, 2 Aug 2003 14:41:56 +0100
   From: "And Rosta" <a.rosta@hidden.email>
Subject: essentials of a gadri system

As a contribution to the discussion, I thought I'd sketch
what I think the essentials of a gadri system would be. I
am ignoring both the current system, and the question of
how to express these essentials verbally.


Accepted. For my part, I have put my contribution on wiki, at
http://www.lojban.org/wiki/index.php/gadri%20report%2C%20aug%202003



In your definitions of intension and extension, where does specificity
enter in? lo mikce is by definition "nonspecific", making it intensional, 
yet it's defined as "da poi broda", which by your explanation is the
archetype of extension!
I think I've just realised why we never agreed on lo != any. Extensional can still be non-specific. In "I saw a doctor", you don't have a name for the doctor in mind, so they aren't specific. But you could meaningfully attach a name to them --- interpret the sentence with a specific reference --- and it would be meaningful; model theoretical semantics works on that very premiss --- you interpret non-specific sentences like "I saw a doctor" by enumerating the possible assignments of a specific value to the referent. In intensional contexts, by contrast, there can be no specific referent plugged in without changing the semantic completely. "I saw Dr Fred" is a subset of "I saw a doctor"; "I looked for Dr Fred" is not a subset of "I looked for a doctor, any doctor". 


Message: 12
   Date: Mon, 4 Aug 2003 12:57:34 +0100
   From: "And Rosta" <a.rosta@hidden.email>
Subject: Re: essentials of a gadri system

John:

And Rosta scripsit:


4. Named Collective. The Collective is identified by name
(cmene). Plus optional noi- and voi-statements (of cardinality
and/or Superset).

Is this like "The Smiths", the rock group whose members are not named
"Smith", or is this like "the Smiths", a family all of whom are named
"Smith"?  I.e. I don't know what "identified by name" means.


It is like the rock group. The plural morpheme in the example is a
distractor. "Fleetwood Mac" would also be an example of (4), as would
"John Cowan".
The proposals treat individuals as singleton collectives. Things don't
have to be done thus, but doing them that way leads to a more minimal
inventory of primitives.
Right. I revert back to saying that the BPFK can't consider redoing gadri from scratch. And CLL p. 124 explicitly rules that lai means the latter -- which precludes the kind of bootstrapping And would prefer. So The Smiths may be loi prenu, but it's la smits., just as it is lo bende, right? 


Message: 13
   Date: Mon, 4 Aug 2003 05:47:47 -0700 (PDT)
   From: Jorge "Llambías" <jjllambias2000@hidden.email>
Subject: Re: Re: essentials of a gadri system


la nitcion cusku di'e


5. Is-a-Subkind-of. Binary predicate taking Kind as one argument.
This is not even a cmavo change (although it could be done with a cmavo 
predicate), and I have no objection to it.


What is a 'cmavo predicate'? Something in GOhA?


As And said, I had MOI in mind.


6. Subset-of: "is n% of" (= "contains x out of every y members
of"), where n can be left vague and have the meaning "is a subset
of". Binary predicates taking 1-4 as one argument.


This can be either a tweak to ce'i, or a related MEI cmavo, possibly
with two numeric arguments, probably not (fractions should do).


Isn't {si'e} for that? {zemuce'isi'e} seems to be "x1 is 75% of x2".


Yes.


7. Each-member-of. -- Distributivizer. Applies to 1-4.


Either this is lu'i, or lu'i converts its referents a single
invididual, whereupon we would likely use a different LAhE.


I hope you meant {lu'a}.


Yes.


Message: 14
   Date: Mon, 4 Aug 2003 09:22:20 -0400 (EDT)
   From: Invent Yourself <xod@hidden.email>
Subject: No static, democratic (was: essentials of a gadri system)

The attribute that distinguishes collectives from pluralities is that a
collective has at least one interesting emergent property, which the
speaker is considering. The null property of belonging to a plurality is 
unhelpful and should not be considered. (Flames piped to /dev/grice.) I
mean instead something like the ability to carry a piano.
As Nick as alluded, de-emphasis of this point is what has seduced so many 
otherwise virtuous lojbanists into using mass gadri simply for plurals.

Nick, please think about adding this point to your wiki exposition.
Done, briefly. I ruminated on this much more in the ontologies; it doesn't really belong in this exposition, because its point was to differentiate loi-collective from loi-stuff rather than from lo-individual. But obviously any complete proposal needs to be explicit about emergence. 


Message: 15
   Date: Mon, 4 Aug 2003 07:04:06 -0700 (PDT)
   From: Jorge "Llambías" <jjllambias2000@hidden.email>
Subject: Re: No static, democratic (was: essentials of a gadri system)

But what And was positing here was to consider the collective
as more basic than the plurality, so one would use collective always
unless you want to emphasize distributiveness. He would use collective
gadri even in the case of a singleton.
Understood. And that's why his proposal is Excellent --- and incompatible with Lojban as we know it. 


It should ceratinly be pointed out that the members of a set have
some properties that apply only to the members one by one, some
properties that apply only to the members as a whole, and some
properties that fall in both camps. I don't think we need to regulate
how to talk about properties that fall in the third category, given
that we are forced to make a choice of talking of them as group or
as individual properties.
I agree that if a property is attributable to both the individuals and the collective, we need not pick one or the other. I don't see the number of such properties being all that big, though I have a nasty suspicion you'll pull some horrid example out... 


Message: 16
   Date: Mon, 4 Aug 2003 15:02:27 +0100
   From: "And Rosta" <a.rosta@hidden.email>
Subject: Re: Re: essentials of a gadri system

Nick:

Accepted. For my part, I have put my contribution on wiki, at
http://www.lojban.org/wiki/index.php/gadri%20report%2C%20aug%202003


We need to know what the inner and outer quantifiers are with the
gadri and LAhE (and what the 'quantifiers' mean in those contexts).
Until then, I can't really evaluate the proposal. I am eager to,
but it is currently too underspecified.
I don't see what's to specify. LAhE1 pisu'o loi ro remna = some human sludge; LAhE2 pisu'o loi ro remna = a crowd. Both, I guess, admit to outer quantifiers counting clumps. Neither admits of an inner integer quantifier. 


I will say, tho, that "LAhE3 su'o lo" will not work. Perhaps "LAhE3
tu'o lo" would work, but that raises the question of what "tu'o lo"
means.
Why? After all, there is no extensional prenex requirement on the referent of LAhE: lo pavyseljirna is non-existent and cannot enter a prenex, but lu'e lo pavyseljirna does exist. So does lu'e ci lo pavyseljirna. If lu'e ci lo pavyseljirna is harmless for "three tokens of 'unicorn' ", why is LAhE3 ci lo pavyseljirna problematic for The Kind Of Three Unicorns, otherwise known as LAhE3 lo pavyseljirna cimei , and the proper expression for "I'm looking for three doctors"? (And looking for 3 unicorns, modulo reification: LAhE3 ci se ka pavyseljirna, perhaps. 


Message: 17
   Date: Mon, 4 Aug 2003 15:28:26 +0100
   From: "And Rosta" <a.rosta@hidden.email>
Subject: Re: Re: essentials of a gadri system

For any property there is a Kind, that is all. In an ideal world, it
would be as easy to talk about the Kind that embodies that property
as about "da" that has the property and "it" that has the property.


The million dollar question, whether {lo se ka} will do it.


4. Named Collective. The Collective is identified by name
(cmene). Plus optional noi- and voi-statements (of cardinality
and/or Superset).


I assume these are (if I can use for now my favoured LAhE-based
solution) LAhE2 ro lo broda, LAhE2 ny lo broda, LAhE2 la broda).


Perhaps: tu'o LAhE2 ro lo broda, tu'o LAhE2 ny lo broda, tu'o
LAhE2 ro la broda.
I know that LAhE1,2,3 could all end up in a new cmavo that doesn't admit outer quantifiers, but why do you rule out outer quantifiers of collectives? You can have two groups called The Smiths. 


5. Is-a-Subkind-of. Binary predicate taking Kind as one argument.
This is not even a cmavo change (although it could be done with a cmavo 
predicate), and I have no objection to it.


We should take it as a given that almost everything can be expressed by
means of full brivla. The utilitarian dimension to the gadri system
involves deciding where to add shortcuts.

In my analysis, "This branch resembles two snakes" is "two subtypes of
Mr Snake", & my judgement is that this is a sufficiently ordinary thing
to want to say that it warrants having a shortcut to say it. In saying
this in a Lojban forum, I am bearing in mind the entire weight of usage
to date, which shows that all users prefer inexactitude to longwindedness, but that Lojbanists do make an effort to learn the ready-packaged devices 
the grammar provides for them. In other words, if the grammar provides
a succinct way to say "two subtypes of", then Lojbanists will learn and
use it. If the grammar doesn't, then Lojbanists will just say "two
snakes" & not worry about saying what they don't mean.
I'm going to need many more examples than the bifurcated snakes to convince myself, let alone others, that this needs a cmavo. If you convince me, though, sure, that's a job for LAhE (this time with an outer quantifier) too. 


6. Subset-of: "is n% of" (= "contains x out of every y members
of"), where n can be left vague and have the meaning "is a subset
of". Binary predicates taking 1-4 as one argument.


This can be either a tweak to ce'i, or a related MEI cmavo, possibly
with two numeric arguments, probably not (fractions should do).


And where does that leave piPA, so'e, "re le ci", & other forms that
belong in the paradigm of (6)?
I'm obviously obtuse here. How are they not merely piPAsi'e, piso'esi'e, refi'ucisi'e...? 


We'd also need a de-broken form of MEI.


Remind me how it's broken again.


7. Each-member-of. -- Distributivizer. Applies to 1-4.


Either this is lu'i, or lu'i converts its referents a single
invididual, whereupon we would likely use a different LAhE. (I'm sure
there are alternatives within the baseline, but they will likely be
cumbersome; e.g. ro lo pafi'uro loi...) We certainly need to settle
which of the two alternatives lu'i means.


We need more details of this, and also of what ro means when applied
to all gadri & LAhE.


Yes, that is on the BPFK list.


8. Existentially quantified variable (Collective). Plus optional
noi/poi-statement of cardinality or Superset.


I don't understand. Is this da-as-collective as distinct from
da-as-individual, or something else?


No. It's ordinary da. IMO the simplest gadri system is achieved by
not distinguishing between Collective and Individual, i.e. by treating
Individual as a singleton Collective. The distinction can be made by
pa versus za'u as a cardinality indicator.
Which is why the Excellent Solutions reject the CLL individual/collective dichotomy.
Remember too that I wasn't saying that the Lojban system should be based 
on these building blocks.
Understood. It needn't be, and it won't be (lo/loi is non-negotiable); but this stuff should be sayable in Lojban, obviously, and should be sayable straightforwardly. 

--
Dr Nick Nicholas           [Stephen] King published _The Green Mile_ as
Research Assistant         the first serialized novel since the 1920s,
French & Italian           in a gesture that was meant to recall the
University of Melbourne    serial work of Dickens. No doubt, King is the
Australia                  Dickens this century deserves.
nickn@hidden.email          -- Richard von Busack, _Metro Santa Cruz_,
http://www.opoudjis.net     Dec. 8-15 1999, p. 29.