Individuals and xorlo

476 views
Skip to first unread message

Dan Rosén

unread,
Feb 3, 2014, 6:36:13 PM2/3/14
to loj...@googlegroups.com

Dear selmriste,

It seems that using xorlo prevents explicitly talking about indivduals, such as
/one elephant/, a seemingly simple concept. Let's start with an inner quantifier:

    lo pa xanto = zo'e noi ke'a xanto gi'e zilkancu li pa lo xanto,

However, the latter {lo xanto} in zilkancu3 can denote about a group of
elephants, so {lo pa xanto} can indeed be many elephants.  Outer quantifiers
will not help, as they will only range over the inner object.

Using zo'e directly is obviously fruitless since xorlo seems to influence how
both zo'e, and how noi work: together they remove our abilities to explicitly
talk about individuals. This make me assume that it also affects the
da-family, so {pa xanto} is also out of the question.

Finally, any brivla will not help us here as the dreaded lo-zo'e-noi-trinity
will always be able to sneak in a group where we want an individual. For
instance in {lo pa kantu be lo pa xanto}, or {lo xantyka'u}, we still might end
up with a onesome of elephants.

Why was it decided to make it like this?  It seems that a monolingual jbopre
would not /really/ be able to differentiate an elephant from its flock.
(But perhaps not if we were talking about sheep, but I digress)

Hopefully I have misunderstood everything. If this is so, please enlighten me.

ki'e mi'e la danr

Felipe Gonçalves Assis

unread,
Feb 3, 2014, 7:14:48 PM2/3/14
to loj...@googlegroups.com
Here is a way if recovering the concept of an individual elephant just from the concepts of elephants and parthood:
A xanto pamei is something that is xanto and that can't be divided in two things such that each one is xanto.

I would express a counting unit with a property:
{lo pa xanto cu zilkancu li pa lo ka xanto}

mu'o
mi'e .asiz.



--
You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+un...@googlegroups.com.
To post to this group, send email to loj...@googlegroups.com.
Visit this group at http://groups.google.com/group/lojban.
For more options, visit https://groups.google.com/groups/opt_out.

Jorge Llambías

unread,
Feb 3, 2014, 7:17:02 PM2/3/14
to loj...@googlegroups.com
On Mon, Feb 3, 2014 at 8:36 PM, Dan Rosén <lur...@gmail.com> wrote:

It seems that using xorlo prevents explicitly talking about indivduals, such as
/one elephant/, a seemingly simple concept. Let's start with an inner quantifier:

    lo pa xanto = zo'e noi ke'a xanto gi'e zilkancu li pa lo xanto,

However, the latter {lo xanto} in zilkancu3 can denote about a group of
elephants, so {lo pa xanto} can indeed be many elephants.  Outer quantifiers
will not help, as they will only range over the inner object.

"lo pa xanto" can only be a single elephant: the elephant in front of you, the African elephant, the elephant being digested by a boa constrictor in Saint-Exupery's drawing, etc. but it always has to be one. It cannot be a group of elephants in front of you, all African elephants, the millions of elephants being digested by boa constrictors in the millions of reproductions of that picture, etc.

Sometimes you can make the same claims about the African elephant that you can make about all African elephants, or about the elephant inside the boa and about all the elephants inside all those boas, or even perhaps about the elephant in front of you and a group of elephants in front of you, but that doesn't mean that linguistically they are the same object.

mu'o mi'e xorxes 

Jorge Llambías

unread,
Feb 3, 2014, 7:21:45 PM2/3/14
to loj...@googlegroups.com
On Mon, Feb 3, 2014 at 9:14 PM, Felipe Gonçalves Assis <felipe...@gmail.com> wrote:
Here is a way if recovering the concept of an individual elephant just from the concepts of elephants and parthood:
A xanto pamei is something that is xanto and that can't be divided in two things such that each one is xanto.

That may (perhaps) work for xanto, but it won't work for djacu. or even worse, for kurfa. 

Felipe Gonçalves Assis

unread,
Feb 3, 2014, 7:32:39 PM2/3/14
to loj...@googlegroups.com

I was pretty sure I was missing something, yes.

With {djacu}, that doesn't create any issues in defining what can be {lo pa djacu}, but with {kurfa}, that means that the concept of kurfa pamei precedes that of kurfa, right? I guess this is what troubles danr.

Dan Rosén

unread,
Feb 4, 2014, 1:31:35 PM2/4/14
to loj...@googlegroups.com
Thank you xorxes and .asiz

"lo pa xanto" can only be a single elephant: the elephant in front of you, the African elephant, the elephant being digested by a boa constrictor in Saint-Exupery's drawing, etc. but it always has to be one. It cannot be a group of elephants in front of you, all African elephants, the millions of elephants being digested by boa constrictors in the millions of reproductions of that picture, etc.

I'm using the expansions suggested in http://www.lojban.org/tiki/BPFK+Section:+gadri, where


    lo pa xanto = zo'e noi ke'a xanto gi'e zilkancu li pa lo xanto,

but {lo xanto} can be plural, so this removes the effect of the zilkancu part. Is it that I misunderstand this equation, or is it just false?

.asiz suggested:

    {lo pa xanto cu zilkancu li pa lo ka xanto},

This might have been better, but the other examples do not use a ka-abstraction in zilkancu3, so zilkancu1 in {zilkancu li pa lo ka xanto}, would seem to be to be a property, not an elephant.

Felipe Gonçalves Assis

unread,
Feb 4, 2014, 4:25:29 PM2/4/14
to loj...@googlegroups.com
On 4 February 2014 16:31, Dan Rosén <lur...@gmail.com> wrote:

    lo pa xanto = zo'e noi ke'a xanto gi'e zilkancu li pa lo xanto,

but {lo xanto} can be plural, so this removes the effect of the zilkancu part. Is it that I misunderstand this equation, or is it just false?


I don't have much to say. I can't make sense of this equation, and disregard it.

Let me just note that making xanto singular wouldn't help me either: Which elephant in the world could possibly be a unit for counting elephants?

Dan Rosén

unread,
Feb 4, 2014, 4:54:17 PM2/4/14
to loj...@googlegroups.com
Let me just note that making xanto singular wouldn't help me either: Which elephant in the world could possibly be a unit for counting elephants?


Ha, good point! ki'e

Jorge Llambías

unread,
Feb 4, 2014, 5:50:52 PM2/4/14
to loj...@googlegroups.com
If you don't think that the elephant is a good unit for counting elephants, you could say "lo gradu be fi lo ka xanto", or, as you said, redefine "kancu" so that it takes a tergradu instead of a gradu in x4.

Do you have the same qualms with "lo mitre" or "lo snidu" to refer to units? What would you fill the x1 of gradu with?

Jorge Llambías

unread,
Feb 4, 2014, 5:56:50 PM2/4/14
to loj...@googlegroups.com
On Tue, Feb 4, 2014 at 3:31 PM, Dan Rosén <lur...@gmail.com> wrote:

I'm using the expansions suggested in http://www.lojban.org/tiki/BPFK+Section:+gadri, where


    lo pa xanto = zo'e noi ke'a xanto gi'e zilkancu li pa lo xanto,

but {lo xanto} can be plural, so this removes the effect of the zilkancu part. Is it that I misunderstand this equation, or is it just false?


I don't think it's quite right to say that "lo xanto" can be plural, because Lojban doesn't have grammatical number, so it can't strictly be singular or plural. But a natural translation of "lo xanto" in this context would indeed be plural in English, something like "is 1 counting in elephants".

Felipe Gonçalves Assis

unread,
Feb 4, 2014, 6:24:43 PM2/4/14
to loj...@googlegroups.com
On 4 February 2014 20:50, Jorge Llambías <jjlla...@gmail.com> wrote:



On Tue, Feb 4, 2014 at 6:25 PM, Felipe Gonçalves Assis <felipe...@gmail.com> wrote:
Let me just note that making xanto singular wouldn't help me either: Which elephant in the world could possibly be a unit for counting elephants?

If you don't think that the elephant is a good unit for counting elephants, you could say "lo gradu be fi lo ka xanto", or, as you said, redefine "kancu" so that it takes a tergradu instead of a gradu in x4.


I am sorry, maybe you can expand on your understanding of {gradu}?

The best I could come up with for using {gradu} is, e.g.,
  {ko'a gradu lo si'o mitre} <=> {ko'a mitre li pa}
  {gradu ko'e ko'i} <=> {ko'e ckilu ko'i}
with, e.g,
  {lo si'o mitre cu ckilu lo ka ma kau ni ce'u clani}
 
Do you have the same qualms with "lo mitre" or "lo snidu" to refer to units? What would you fill the x1 of gradu with?


As exposed, I would indeed fill the x1 of {gradu} with {lo mitre} and {lo snidu}, but to me these are references to things that are one meter and one second long, respectively. They are concrete stuff. A standard of measurement would be a se gradu or a ckilu.

Jorge Llambías

unread,
Feb 4, 2014, 6:49:29 PM2/4/14
to loj...@googlegroups.com
On Tue, Feb 4, 2014 at 8:24 PM, Felipe Gonçalves Assis <felipe...@gmail.com> wrote:

I am sorry, maybe you can expand on your understanding of {gradu}?

My personal "definition" of gradu is this:

...
milti
centi
decti 
gradu
dekto
xecto
kilto
...

same type of place structure for all of them. 


The best I could come up with for using {gradu} is, e.g.,
  {ko'a gradu lo si'o mitre} <=> {ko'a mitre li pa}
  {gradu ko'e ko'i} <=> {ko'e ckilu ko'i}
with, e.g,
  {lo si'o mitre cu ckilu lo ka ma kau ni ce'u clani}

I never really understood the connection of "si'o" with scales. 

Felipe Gonçalves Assis

unread,
Feb 4, 2014, 7:02:20 PM2/4/14
to loj...@googlegroups.com
On 4 February 2014 21:49, Jorge Llambías <jjlla...@gmail.com> wrote:


On Tue, Feb 4, 2014 at 8:24 PM, Felipe Gonçalves Assis <felipe...@gmail.com> wrote:

I am sorry, maybe you can expand on your understanding of {gradu}?

My personal "definition" of gradu is this:

...
milti
centi
decti 
gradu
dekto
xecto
kilto
...

same type of place structure for all of them. 


How would you contrast {gradu} and {dunli}?
 

The best I could come up with for using {gradu} is, e.g.,
  {ko'a gradu lo si'o mitre} <=> {ko'a mitre li pa}
  {gradu ko'e ko'i} <=> {ko'e ckilu ko'i}
with, e.g,
  {lo si'o mitre cu ckilu lo ka ma kau ni ce'u clani}

I never really understood the connection of "si'o" with scales. 


I am just using {si'o} as a generic relationship abstraction here. I guess it would be
{lo ka ce'u mitre ma kau}.

Jorge Llambías

unread,
Feb 4, 2014, 7:18:54 PM2/4/14
to loj...@googlegroups.com
On Tue, Feb 4, 2014 at 9:02 PM, Felipe Gonçalves Assis <felipe...@gmail.com> wrote:

How would you contrast {gradu} and {dunli}?


I think "mi dunli do lo ka ce'u citka ma kau" is normal, whereas "mi gradu do ..." is bizarre.

Felipe Gonçalves Assis

unread,
Feb 4, 2014, 7:48:55 PM2/4/14
to loj...@googlegroups.com
Sure! va'i {gradu} is about scales stricto sensu.

Well, as to how this affects the previous exchange, using your definition of {gradu}, I would say that a standard of measurement could be a ka gradu.

Gleki Arxokuna

unread,
Feb 5, 2014, 2:53:25 AM2/5/14
to loj...@googlegroups.com
Sorry for intruding. I need to explain this in simple words for a future lojban tutorial.
So 
{zo'e} denotes an individual/individuals.
{lo najgenja} = carrot/carrots
{ci lo najgenja cu grake li 60} = {ci zo'e noi najgenja cu grake li 60} - describes carrots. Three of carrots are 60 grams each.

Now I postulate an axiom that {[su'o] lo pa najgenja} describes one carrot (I'll avoid formulae here since i need it for a tutorial, not for a reference grammar).
{ro lo ci najgenja} describes each of the three carrots.

Two important conclusions:
1. {ro lo ci najgenja cu grake li 60} - one carrot is always 60 grams in weight.
2. {ro loi ci najgenja cu grake li 60} = {ro zo'e noi gunma lo ci najgenja cu grake li 60} - describes masses (again of carrots but carrots here are of less importance since carrots are hidden inside gunma2). Each mass of carrots (with three carrots in each mass) is 60 grams so each carrots weighs 20 grams on average.

Is my reasoning correct?
I remember someone saying that {lo} is more vague and might include masses as well but here {loi} and it's underlying {gunma} move carrots higher. Can we accept raising here? If yes then all this reasoning immediately breaks.




--

selpa'i

unread,
Feb 5, 2014, 6:40:03 AM2/5/14
to loj...@googlegroups.com
la .xorxes. cu cusku di'e
> {lo si'o mitre cu ckilu lo ka ma kau ni ce'u clani}
>
> I never really understood the connection of "si'o" with scales.

Me neither. I don't understand how it makes sense, as it's very
difficult to interact with a {si'o} systematically. I use {ka} instead,
sometimes with two {ce'u}, although I find the {ce'u ma kau} way just as
intuitive, for example in klani3:

lo vi rokci cu klani li mu lo ka ce'u ki'ogra ce'u/ma kau

which means the same as

lo vi rokci cu ki'ogra li mu

How to use {klani} with predicates that don't have an obvious
input-output pair is another question. Here, {kau} seems to be more
convenient as the output marker than a second {ce'u}, say:

mi klani li ci lo ka ce'u citka xo kau plise
"I measure three in how many apples I ate."

Probably the more difficult challenge is to find a situation where klani
is actually the more practical (or the only practical) option. Maybe {ni}:

li vo ni mi bajysru lo foldi [kei lo ka ce'u xo kau roi fasnu]
"Four is the quantity of my running around the the field, [measured
in how many times it happened]"

But then, that's still only a complicated way of saying:

mi vo roi bajysru lo foldi

So... the challenge stands.

Though maybe the advantage lies in the vagueness of klani3-less
{klani}-usage. I'm really not sure.

mi'e la selpa'i mu'o

selpa'i

unread,
Feb 5, 2014, 6:47:54 AM2/5/14
to loj...@googlegroups.com
la .xorxes. cu cusku di'e
If I may, Dan is asking why the unit {lo xanto} cannot be (implicitly)
{lo ci xanto}, in which case three elephants would be counted as one
counting off by threes. Using a property in zilkancu3 would probably be
clearer for that reason. As it stands, some people seem to think that
the zilkancu3 unit contains a context-dependent inner quantifier, thus
counting of by {xo'e mei}. I don't think that's the intended meaning, so
it should be stated clearly that we're dealing with singletons.

selpa'i

unread,
Feb 5, 2014, 11:49:34 AM2/5/14
to loj...@googlegroups.com
la .dan. cu cusku di'e
> Using zo'e directly is obviously fruitless since xorlo seems to
> influence how both zo'e, and how noi work: together they remove our abilities to
> explicitly talk about individuals.

I don't think {noi} changed at all. {zo'e} allows plural reference, but
that isn't new either.

> This make me assume that it also affects the
> da-family, so {pa xanto} is also out of the question.

This comes down to whether or not {da} allows for plural variables.
Since plural reference is so common in Lojban, it would make sense for
{da} to also allow plural variables, but singular variables also have
advantages.

Imagine {za'a lo ci xanto cu va cadzu} to set the context. Now, it all
depends on on {da}'s plurality what {da va cadzu} can mean. Clearly, we
just saw that {lo ci xanto} is a cadzu1, so it should be a possible
value for the {da}. The downside to this is that with plural variables,
the one X in {pa lo ci xanto} could be all three elephants (although a
distributive handling of {me}'s x1 could fix that, or in other words, by
saying that {mi'o na me mi'o}), whereas singular variables could only
pick out an individual elephant from {lo ci xanto}.

So singular variables are simpler and avoid certain problems, like the
{pa xanto} one. On the other hand, it would mean that we can't say {da
simxu lo ka prami} for "There are some X who love each other", and we'd
have to use more complicated mechanisms for that, like {da poi su'o mei
cu simxu lo ka prami} (which isn't *that* bad).

Personally I would be all but opposed to the idea of having plural
variables to along with the plural reference while keeping the
simplicity of singular quantification, but I probably can't have my cake
and eat it, too. I would not want two sets of quantifers, for eaxmple.
Another idea would be to have each selbri place decide if it's
distributive or not, but I'm not sure I like that very much. So the more
practical solution right now seems to be to stick with singular
variables, even though it breaks the {simxu} example above and can
sometimes be counter-intuitive in a language full of plural reference.

John E Clifford

unread,
Feb 5, 2014, 1:19:31 PM2/5/14
to loj...@googlegroups.com
It seems that the constantly evolving xorlo (now moving far from 'lo' -- and probably from xorxes as well) has gotten itself into yet another jam, presumably from trying to do too much again.  I suspect that what is needed is to go back to basics and get that clear once more and then move ahead cautiously.  So, the basic 'lo broda' is "the salient node of the upward semilattice of jest on the set assigned to 'broda'" (some set of brodas and broda parts -- whatever that may mean for a particular kind of thing as broda).  'lo', unlike 'loi' says nothing about how the set involved is connected to the predicates involved (collective or distributive).  Variables range over L-sets or are plural, depending on your mathematical theology.  Etc.  do we need to fill in all the details and, if not, which ones?

From: selpa'i <sel...@gmx.de>
To: loj...@googlegroups.com
Sent: Wednesday, February 5, 2014 10:49 AM
Subject: Re: [lojban] Individuals and xorlo
-- You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+unsub...@googlegroups.com.

selpa'i

unread,
Feb 5, 2014, 2:05:52 PM2/5/14
to loj...@googlegroups.com
la .pycyn. cu cusku di'e
> It seems that the constantly evolving xorlo (now moving far from 'lo' --
> and probably from xorxes as well)

I don't think so. It's just that some questions that often get asked
about xorlo are actually more general than xorlo. Nothing is changing
about xorlo either, it's just that when new people come around, they
find a mess of underdocumented and scattered definitions and need to ask
the same (or similar) questions each time again. This will only stop
once there is a more complete specification that one can point someone
to instead of having to re-open a discussion about xorlo.

> So, the basic 'lo broda' is "the salient node of
> the upward semilattice of jest on the set assigned to 'broda'" (some set
> of brodas and broda parts -- whatever that may mean for a particular
> kind of thing as broda). 'lo', unlike 'loi' says nothing about how the
> set involved is connected to the predicates involved (collective or
> distributive).

All that is well-known, although we tend to use different terminology.

> Variables range over L-sets or are plural, depending on
> your mathematical theology. Etc. do we need to fill in all the details
> and, if not, which ones?

We need to know if variables are plural or singular, that's all.
Currently, they are "defined" as singular, for some value of "defined"
that makes sense when there is no official body to define it.

The rules are simple enough, they just aren't layed out properly. (If
they were, we wouldn't be having this discussion.)

guskant

unread,
Feb 5, 2014, 3:09:54 PM2/5/14
to loj...@googlegroups.com


Le mercredi 5 février 2014 20:47:54 UTC+9, selpa'i a écrit :
la .xorxes. cu cusku di'e 
> On Tue, Feb 4, 2014 at 3:31 PM, Dan Rosén <lur...@gmail.com 
> <mailto:lur...@gmail.com>> wrote: 


>     I'm using the expansions suggested in 
>     http://www.lojban.org/tiki/BPFK+Section:+gadri, where 


>          lo pa xanto = zo'e noi ke'a xanto gi'e zilkancu li pa lo xanto, 

>     but {lo xanto} can be plural, so this removes the effect of the 
>     zilkancu part. Is it that I misunderstand this equation, or is it 
>     just false? 


> I don't think it's quite right to say that "lo xanto" can be plural, 
> because Lojban doesn't have grammatical number, so it can't strictly be 
> singular or plural. But a natural translation of "lo xanto" in this 
> context would indeed be plural in English, something like "is 1 counting 
> in elephants". 

If I may, Dan is asking why the unit {lo xanto} cannot be (implicitly) 
{lo ci xanto}, in which case three elephants would be counted as one 
counting off by threes. Using a property in zilkancu3 would probably be 
clearer for that reason. As it stands, some people seem to think that 
the zilkancu3 unit contains a context-dependent inner quantifier, thus 
counting of by {xo'e mei}. I don't think that's the intended meaning, so 
it should be stated clearly that we're dealing with singletons. 


If you mean simply "one-some" of a mass with the word "singleton", I agree with you for English "explanation" of {lo PA broda}. As for Lojban "definition", I would rather support the current definition, and need a Lojban definition of {kancu}, which is used in the definition of {zilkancu}.

I suggest as follows:

{x1 kancu x2 x3 x4} =ca'e {gau x1 boi x2 se tcita x3 noi ke'a namcu gi'e x3 mei x4 noi ke'a gradu}

I'm not sure if this definition would be totally reasonable, but at least it mentions {x4 noi ke'a gradu}, consequently {x4 pa mei} because of definition of {gradu}={x1 pa mei gi'e ckaji x3 noi se ckilu x2}.
With this definition, {zilkancu}_3 is clearly defined as {pamei}_1, and no other explanation is necessary.


However, if you mean "individual" with the word "singleton", it is better not to state it, because any mass, no matter if it is used as collective or distributive, can be a unit "one-some" in some sense.
An individual is defined as follows (based on Plural Predication by Thomas McKay, 2006):

"SUMTI is individual" =ca'e {RO DA poi ke'a me SUMTI zo'u SUMTI me DA}
where RO and DA are not a singular quantifier {ro} and a singular variable {da} of Lojban, but a plural quantifier and a plural variable respectively.

If {zilkancu}_3 should be always an individual, {lo ckafi} is not an individual in many cases of universe of discourse, and it cannot be {zilkancu}_3.
However, {lo ckafi} can be naturally a unit:
{mi cpedu tu'a lo pa ckafi} = {mi cpedu tu'a zo'e noi ke'a ckafi gi'e zilkancu li pa lo ckafi}

This flexibility of {zilkancu}_3, the unit, is advantage of xorlo, and indispensable for keeping expressiveness of Lojban.

 
Le jeudi 6 février 2014 01:49:34 UTC+9, selpa'i a écrit :

So singular variables are simpler and avoid certain problems, like the
{pa xanto} one. On the other hand, it would mean that we can't say {da
simxu lo ka prami} for "There are some X who love each other", and we'd
have to use more complicated mechanisms for that, like {da poi su'o mei
cu simxu lo ka prami} (which isn't *that* bad).


No, because the domain of {da} of {da poi (ke'a) su'o (re) mei} spans distributively over plural {su'o (re) mei}_1. 
{da poi ke'a gunma cu simxu lo ka prami} treats the plural {simxu}_1 collectively,
just like a developed form of {su'o loi}={su'o da poi ke'a me lo gunma be lo}.

 


Le mercredi 5 février 2014 16:53:25 UTC+9, la gleki a écrit :
Sorry for intruding. I need to explain this in simple words for a future lojban tutorial.
So 
{zo'e} denotes an individual/individuals.
{lo najgenja} = carrot/carrots
{ci lo najgenja cu grake li 60} = {ci zo'e noi najgenja cu grake li 60} - describes carrots. Three of carrots are 60 grams each.

Now I postulate an axiom that {[su'o] lo pa najgenja} describes one carrot (I'll avoid formulae here since i need it for a tutorial, not for a reference grammar).
{ro lo ci najgenja} describes each of the three carrots.

Two important conclusions:
1. {ro lo ci najgenja cu grake li 60} - one carrot is always 60 grams in weight.
2. {ro loi ci najgenja cu grake li 60} = {ro zo'e noi gunma lo ci najgenja cu grake li 60} - describes masses (again of carrots but carrots here are of less importance since carrots are hidden inside gunma2). Each mass of carrots (with three carrots in each mass) is 60 grams so each carrots weighs 20 grams on average.
 
Is my reasoning correct?


Yes, it is correct.

 
I remember someone saying that {lo} is more vague and might include masses as well but here {loi} and it's underlying {gunma} move carrots higher. Can we accept raising here? If yes then all this reasoning immediately breaks.



{lo} can be a mass, but it does not say if the mass satisfies the predicate collectively or/and distributively.
On the other hand, {loi}={lo gunma be lo} says that the mass satisfies the predicate collectively.

When an outer PA is attached to the sumti, the implicit {da} spans distributively over the domain:
{ro lo ci najgenja}={ro da poi me lo ci najgenja}
{ro loi ci najgenja}={ro da poi me lo gunma be lo ci najgenja}
You see, the domain of {ro lo ci najgenja} is each {najgenja}, and that of {ro loi ci najgenja} is each {gunma}.


 

Jorge Llambías

unread,
Feb 5, 2014, 3:26:27 PM2/5/14
to loj...@googlegroups.com
On Wed, Feb 5, 2014 at 8:47 AM, selpa'i <sel...@gmx.de> wrote:

         lo pa xanto = zo'e noi ke'a xanto gi'e zilkancu li pa lo xanto,

If I may, Dan is asking why the unit {lo xanto} cannot be (implicitly) {lo ci xanto}, in which case three elephants would be counted as one counting off by threes. Using a property in zilkancu3 would probably be clearer for that reason.

 But "lo ka ce'u xanto" is a property of "lo ci xanto" as well as of "lo xanto", so I'm not sure it adds anything in that respect. "ka" does help people who don't like generic references, but I'm not sure it does anything more than "lo xanto" to specify the size of the unit. How would it help to say that you are counting by things that have the property "lo ka ce'u xanto" if among those things there's lo ci xanto as well as lo pa xanto?

As it stands, some people seem to think that the zilkancu3 unit contains a context-dependent inner quantifier, thus counting of by {xo'e mei}. I don't think that's the intended meaning, so it should be stated clearly that we're dealing with singletons.
 
An important point of xorlo is that it gets rid of implicit quantifiers, but yes.  

selpa'i

unread,
Feb 5, 2014, 3:28:08 PM2/5/14
to loj...@googlegroups.com
la .guskant. cu cusku di'e
> Le mercredi 5 février 2014 20:47:54 UTC+9, selpa'i a écrit :
> If I may, Dan is asking why the unit {lo xanto} cannot be (implicitly)
> {lo ci xanto}, in which case three elephants would be counted as one
> counting off by threes. Using a property in zilkancu3 would probably be
> clearer for that reason. As it stands, some people seem to think that
> the zilkancu3 unit contains a context-dependent inner quantifier, thus
> counting of by {xo'e mei}. I don't think that's the intended
> meaning, so
> it should be stated clearly that we're dealing with singletons.
>
>
> If you mean simply "one-some" of a mass with the word "singleton", I
> agree with you for English "explanation" of {lo PA broda}. As for Lojban
> "definition", I would rather support the current definition, and need a
> Lojban definition of {kancu}, which is used in the definition of {zilkancu}.

Right, I'm not proposing to change the definition. I only explained the
reason for Dan's confusion. Making zilkancu (or kancu) clearer, would
solve the problem, but it would also help to explicitly state (in
English, for beginners) that in {lo PA broda}, we don't count by context
dependent units. Counting off by {lo broda} is intended to mean that {lo
ci broda} contains three individuals that each {broda}. This is what the
current definitions tries to say. It just wasn't clear enough for Dan or
la latro'a.

> However, if you mean "individual" with the word "singleton", it is
> better not to state it, because any mass, no matter if it is used as
> collective or distributive, can be a unit "one-some" in some sense.

Once you have a mass, then that mass is a new individual altogether. But
a sumti like {mi'o} or {mi jo'u do} is not a mass, it's just two
individuals together.

> An individual is defined as follows (based on Plural Predication by
> Thomas McKay, 2006):
>
> "SUMTI is individual" =ca'e {RO DA poi ke'a me SUMTI zo'u SUMTI me DA}
> where RO and DA are not a singular quantifier {ro} and a singular
> variable {da} of Lojban, but a plural quantifier and a plural variable
> respectively.

Yes, that is exactly the definition of "individual" I am using.

> If {zilkancu}_3 should be always an individual, {lo ckafi} is not an
> individual in many cases of universe of discourse, and it cannot be
> {zilkancu}_3.

{lo ckafi} is an amount of coffee. If I have two separate amounts of
coffee, then I can count them together {lo re ckafi}.

I would still call {lo ckafi} an individual. Using a property in
zilkancu3 has been suggested, so we either count by {lo ckafi} or {lo ka
ckafi}. The thing that makes {lo pa ckafi} different from {lo pa prenu}
is that splitting {lo pa ckafi} will result in two new {lo ckafi},
whereas splitting a person will just... kill it.

> However, {lo ckafi} can be naturally a unit:
> {mi cpedu tu'a lo pa ckafi} = {mi cpedu tu'a zo'e noi ke'a ckafi gi'e
> zilkancu li pa lo ckafi}

Certainly.

> This flexibility of {zilkancu}_3, the unit, is advantage of xorlo, and
> indispensable for keeping expressiveness of Lojban.

I don't think anyone is trying to remove flexible units.

selpa'i

unread,
Feb 5, 2014, 4:00:10 PM2/5/14
to loj...@googlegroups.com
la .xorxes. cu cusku di'e
> Using a property in zilkancu3
> would probably be clearer for that reason.
>
>
> But "lo ka ce'u xanto" is a property of "lo ci xanto" as well as of
> "lo xanto", so I'm not sure it adds anything in that respect. "ka" does
> help people who don't like generic references,

Personally, I don't mind generic references.

> but I'm not sure it does
> anything more than "lo xanto" to specify the size of the unit. How would
> it help to say that you are counting by things that have the property
> "lo ka ce'u xanto" if among those things there's lo ci xanto as well as
> lo pa xanto?

You're right, it doesn't help at all, because {lo ci xanto cu xanto}.
"to be an elephant" is a distributive predicate, but Lojban doesn't put
that job on the selbri. We could try:

lo ci xanto cu xanto gi'e zilkancu li ci lo ka ro xo kau ce'u xanto

And now I'd rather go back to {lo xanto} as the unit. :)

MorphemeAddict

unread,
Feb 5, 2014, 5:11:51 PM2/5/14
to loj...@googlegroups.com
It seems to me that if you allow "pa xanto" or "pa lo xanto" to mean anything other than "one elephant" without relying on context, there's going to be a problem. 

stevo


--
You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+unsubscribe@googlegroups.com.

Jonathan Jones

unread,
Feb 5, 2014, 6:35:50 PM2/5/14
to loj...@googlegroups.com
On Wed, Feb 5, 2014 at 12:05 PM, selpa'i <sel...@gmx.de> wrote:
la .pycyn. cu cusku di'e
<snip>
underdocumented and scattered definitions.... [M]ore complete specification that one can point someone to instead of having to re-open a discussion about xorlo.

The official definitions of cmavo are organized by selma'o on the BPFK Sections page. This page should allow adding such relevant information. I would suggest either the general Notes or Issues, or specific Notes for the gadri itself.

I would assume these pages also exist on the MediaWiki, but you'd have to ask the maintainer, Gleki, where they are.

<snip>

Variables range over L-sets or are plural, depending on
your mathematical theology.  Etc.  do we need to fill in all the details
and, if not, which ones?

We need to know if variables are plural or singular, that's all. Currently, they are "defined" as singular, for some value of "defined" that makes sense when there is no official body to define it.

Sets in general can be any number and are neither singular nor plural. It is possible to have 0-sets, 1-sets, or even ∞-sets. A specific set will be either singular, plural, or neither depending entirely on how many members it has in it. In Lojban the precise set being referenced (, i.e., the members of the set {la.djan.} are all entities who have "John" or its variant spellings as the name by which they are called), can usually be inferred from context (, i.e., those named "John" who are relevant to the context of the current discussion, which in this case is John Clifford and myself.)

In the case of {lo ve kalcu}, I would say that the set would always, or at least nearly always be a 1-set, and would typically be the 1-set [1], although I can imagine counting by multiples, such as in {kancu li cire to du li ci te'a mu toi me'o li mu li re}.
 
<snip>



--
mu'o mi'e .aionys.

.i.e'ucai ko cmima lo pilno be denpa bu .i doi.luk. mi patfu do zo'o
(Come to the Dot Side! Luke, I am your father. :D )

Jonathan Jones

unread,
Feb 5, 2014, 6:47:50 PM2/5/14
to loj...@googlegroups.com
sa du li re te'a mu toi li mu me'o pi'i re

guskant

unread,
Feb 5, 2014, 11:34:34 PM2/5/14
to loj...@googlegroups.com


Le jeudi 6 février 2014 05:28:08 UTC+9, selpa'i a écrit :
la .guskant. cu cusku di'e
> Le mercredi 5 février 2014 20:47:54 UTC+9, selpa'i a écrit :
>     If I may, Dan is asking why the unit {lo xanto} cannot be (implicitly)
>     {lo ci xanto}, in which case three elephants would be counted as one
>     counting off by threes. Using a property in zilkancu3 would probably be
>     clearer for that reason. As it stands, some people seem to think that
>     the zilkancu3 unit contains a context-dependent inner quantifier, thus
>     counting of by {xo'e mei}. I don't think that's the intended
>     meaning, so
>     it should be stated clearly that we're dealing with singletons.
>
>
> If you mean simply "one-some" of a mass with the word "singleton", I
> agree with you for English "explanation" of {lo PA broda}. As for Lojban
> "definition", I would rather support the current definition, and need a
> Lojban definition of {kancu}, which is used in the definition of {zilkancu}.

Right, I'm not proposing to change the definition. I only explained the
reason for Dan's confusion. Making zilkancu (or kancu) clearer, would
solve the problem, but it would also help to explicitly state (in
English, for beginners) that in {lo PA broda}, we don't count by context
dependent units. Counting off by {lo broda} is intended to mean that {lo
ci broda} contains three individuals that each {broda}. This is what the
current definitions tries to say. It just wasn't clear enough for Dan or
la latro'a.



That's nice. 
Although it will become out of topic, I have another suggestion related to the BPFK page of gadri.

"Any term without an explicit outer quantifier is a constant" should be changed to 
"Any term without an explicit outer quantifier can be a constant",
because an usual predicate logic has an axiom on a constant c that "F(c) {inaja} there is at least one (individual) x such that F(x)";
this means that the sentence "any term without an explicit outer quantifier is a constant" automatically implicates an outer quantifier {su'o},
and it contradicts to xorlo itself that there are no default quantifiers.

Most general term, without quantifier, with no universe of discourse yet defined, should be called "free variable".
Once a context is given, it defines an universe of discourse, then each free variable in a sentence becomes a bound plural variable OR a constant (not always a constant), then the truth value of the sentence is specified; if a term denotes an individual, it can become a bound singular variable, then an outer quantifier of Lojban is also available for the term.
The whole procedure depends on the context, and the language itself should not define that a term is a constant.


 
> However, if you mean "individual" with the word "singleton", it is
> better not to state it, because any mass, no matter if it is used as
> collective or distributive, can be a unit "one-some" in some sense.

Once you have a mass, then that mass is a new individual altogether. But
a sumti like {mi'o} or {mi jo'u do} is not a mass, it's just two
individuals together.



I use the term "mass" as something in a domain of plural variable, saying nothing about collectivity/distributivity.
I know BPFK and you use the term "mass" only for "collective mass", but I think this usage is confusing for beginners, because:

1. CLL uses the term "mass" more generally, not always for collective mass;
2. the English word "mass" is too vague to be used as a technical term that involving collectivity;
3. it is useful to define "mass" as follows:
"mass" =ca'e "something in a domain of plural variable";
"collective mass" =ca'e "mass that satisfies the predicate collectively";
"distributive mass" =ca'e "mass that satisfies the predicate distributively".

If you suggest another short term for "something in a domain of plural variable, saying nothing about collectivity/distributivity", I would abandon my usage of "mass" in this meaning.



 
> An individual is defined as follows (based on Plural Predication by
> Thomas McKay, 2006):
>
> "SUMTI is individual" =ca'e {RO DA poi ke'a me SUMTI zo'u SUMTI me DA}
> where RO and DA are not a singular quantifier {ro} and a singular
> variable {da} of Lojban, but a plural quantifier and a plural variable
> respectively.

Yes, that is exactly the definition of "individual" I am using.

> If {zilkancu}_3 should be always an individual, {lo ckafi} is not an
> individual in many cases of universe of discourse, and it cannot be
> {zilkancu}_3.

{lo ckafi} is an amount of coffee. If I have two separate amounts of
coffee, then I can count them together {lo re ckafi}.

I would still call {lo ckafi} an individual. Using a property in
zilkancu3 has been suggested, so we either count by {lo ckafi} or {lo ka
ckafi}. The thing that makes {lo pa ckafi} different from {lo pa prenu}
is that splitting {lo pa ckafi} will result in two new {lo ckafi},
whereas splitting a person will just... kill it.


Yes, but whether {lo ckafi}, {lo prenu} etc. are individual or not depends on epistemology, and the epistemology depends on the universe of discourse, on the context.
It is not defined by Lojban.
 

Gleki Arxokuna

unread,
Feb 6, 2014, 1:08:38 AM2/6/14
to loj...@googlegroups.com

On Thu, Feb 6, 2014 at 3:35 AM, Jonathan Jones <eye...@gmail.com> wrote:
I would assume these pages also exist on the MediaWiki, but you'd have to ask the maintainer, Gleki, where they are.



However, maintainer is supposed to maintain a wiki in a working state. Why asking the maintainer of Wikipedia where this or that information exists?

However, since it's not an official wiki the question for now is only to have the mediawiki in sync with the official wiki. BPFK sections are the most important part so it's better to always read the tiki.

selpa'i

unread,
Feb 6, 2014, 6:33:19 AM2/6/14
to loj...@googlegroups.com
la .guskant. cu cusku di'e
> Once you have a mass, then that mass is a new individual altogether.
> But
> a sumti like {mi'o} or {mi jo'u do} is not a mass, it's just two
> individuals together.
>
> I use the term "mass" as something in a domain of plural variable,
> saying nothing about collectivity/distributivity.
> I know BPFK and you use the term "mass" only for "collective mass", but
> I think this usage is confusing for beginners, because:
>
> 1. CLL uses the term "mass" more generally, not always for collective mass;
> 2. the English word "mass" is too vague to be used as a technical term
> that involving collectivity;
> 3. it is useful to define "mass" as follows:
> "mass" =ca'e "something in a domain of plural variable";
> "collective mass" =ca'e "mass that satisfies the predicate collectively";
> "distributive mass" =ca'e "mass that satisfies the predicate
> distributively".

The term "mass" is confusing exactly because it has been used to mean so
many different things. I would avoid the term myself. However, whenever
I say mass, I mean {gunma}.

{lo gunma} is an individual, too. The referent of {lo gunma} is the
"mass", not its members, which is the whole point of {gunma}. I also
think that {gunma}'s semantics aren't very clear. We still don't have a
definite answer on what properties a {gunma} has, how those properties
are related to its members, and whether it can attain new properties,
and which ones. For me, a {gunma} is a whole new entity, and it can be
the value of a singular variable. There is also no question of
distributivity with {gunma}, as it is just one thing (unless you have
multiple {gunma}, in which case {lo PA gunma} is the same as any other
{lo PA broda}, not specifying distributivity).

> If you suggest another short term for "something in a domain of plural
> variable, saying nothing about collectivity/distributivity", I would
> abandon my usage of "mass" in this meaning.

A term that I've been using, but which doesn't seem to be very
wide-spread (yet?), is "individual-collection". Anything that can be
expressed as {X jo'u Y jo'u Z ...} is an individual collection and is
identical to a {lo broda} with those {jo'u}-connected referents.

> Yes, but whether {lo ckafi}, {lo prenu} etc. are individual or not
> depends on epistemology, and the epistemology depends on the universe of
> discourse, on the context.
> It is not defined by Lojban.

The way I see it, any {lo broda} is an individual (or an
individual-collection). It doesn't matter what {broda} is. What kind of
individuals there are in {lo broda} depends on {broda}, but they are
still always individuals. There is no difference between {lo ckafi} and
{lo prenu} in terms of individualness.

Jorge Llambías

unread,
Feb 6, 2014, 4:22:09 PM2/6/14
to loj...@googlegroups.com
On Thu, Feb 6, 2014 at 1:34 AM, guskant <gusni...@gmail.com> wrote:

Although it will become out of topic, I have another suggestion related to the BPFK page of gadri.

"Any term without an explicit outer quantifier is a constant" should be changed to 
"Any term without an explicit outer quantifier can be a constant",
because an usual predicate logic has an axiom on a constant c that "F(c) {inaja} there is at least one (individual) x such that F(x)";

That applies to singular constants, whereas unquantified terms need not be singular, but the version with plural quantifiers will still be valid.
 
this means that the sentence "any term without an explicit outer quantifier is a constant" automatically implicates an outer quantifier {su'o},

It shouldn't implicate that. "F{c} -> Ex F(x)" does not mean that "F(c)" and "Ex F(x)" have the same meaning, nor that "c" is just a shorthand for "Ex ...x...". Similarly xorlo says that "lo broda" is not just shorthand for "su'o lo broda".
 
and it contradicts to xorlo itself that there are no default quantifiers.

Not just no default quantifiers. No implicit hidden quantifiers at all, The point is that "lo broda" is not a quantification of the bridi it appears in, the way "su'o lo broda" is.

Jorge Llambías

unread,
Feb 6, 2014, 4:33:37 PM2/6/14
to loj...@googlegroups.com
On Thu, Feb 6, 2014 at 8:33 AM, selpa'i <sel...@gmx.de> wrote:
la .guskant. cu cusku di'e

If you suggest another short term for "something in a domain of plural
variable, saying nothing about collectivity/distributivity", I would
abandon my usage of "mass" in this meaning.

A term that I've been using, but which doesn't seem to be very wide-spread (yet?), is "individual-collection". Anything that can be expressed as {X jo'u Y jo'u Z ...} is an individual collection and is identical to a {lo broda} with those {jo'u}-connected referents.

The problem with that, with whatever term is chosen in the metalanguage to talk about the language, is that the term eventually leaks into the language, and then collections become individuals too. That happened to "mass" very quickly, probably from the start, compounded with the problem that it was used for a lot of other things and not just for plural reference. I prefer to say that the domain of plural variables is just individuals, not something else, but that variables don't need to take one value at a time.

selpa'i

unread,
Feb 6, 2014, 4:59:50 PM2/6/14
to loj...@googlegroups.com
la .xorxes. cu cusku di'e
> On Thu, Feb 6, 2014 at 8:33 AM, selpa'i <sel...@gmx.de
> <mailto:sel...@gmx.de>> wrote:
>
> la .guskant. cu cusku di'e
>
>
> If you suggest another short term for "something in a domain of
> plural
> variable, saying nothing about collectivity/distributivity", I would
> abandon my usage of "mass" in this meaning.
>
>
> A term that I've been using, but which doesn't seem to be very
> wide-spread (yet?), is "individual-collection". Anything that can be
> expressed as {X jo'u Y jo'u Z ...} is an individual collection and
> is identical to a {lo broda} with those {jo'u}-connected referents.
>
>
> The problem with that, with whatever term is chosen in the metalanguage
> to talk about the language, is that the term eventually leaks into the
> language, and then collections become individuals too.

That is indeed a possible danger, I agree. But of course we need to be
able to talk about it *somehow* (both in English and in Lojban),
although I will grant that it is logic/linguistics jargon and thus of
lower priority than everyday life vocabulary, which Lojban is still
lacking to a great extent.

> I prefer to say that the domain of plural variables is
> just individuals, not something else, but that variables don't need to
> take one value at a time.

I agree 100% with that.

jacfold...@gmail.com

unread,
Feb 6, 2014, 8:52:48 PM2/6/14
to loj...@googlegroups.com
Do you still mean 
"SUMTI is individual" =ca'e {RO DA poi ke'a me SUMTI zo'u SUMTI me DA}
with the term "individual"?

If so, keeping {lo broda} to be individual requires attentiveness on the universe of discourse, and reduces the flexibility of the language.

Let me give an example.

lo prenu cu jmaji gi'e jukpa gi'e citka

I want to mean with this sentence that this {lo prenu} consists of at least two persons {by} and {cy}, and satisfies {jmaji} collectively {je} non-distributively, {jukpa} collectively {ja} distributively, {citka} non-collectively {je} distributively. This {lo} cannot be replaced by {loi} because I want it to satisfy a selbri non-collectively.

For this {lo prenu}, {by me lo prenu} is true, but {lo prenu me by} is false, so this {lo prenu} is not individual.
It is still possible that you don't include {by} and {cy} in your universe of discourse and say that {lo prenu} is individual, but it is your epistemology, and not defined by the language.

 

guskant

unread,
Feb 6, 2014, 8:54:57 PM2/6/14
to loj...@googlegroups.com
Do you still mean 
"SUMTI is individual" =ca'e {RO DA poi ke'a me SUMTI zo'u SUMTI me DA}

guskant

unread,
Feb 6, 2014, 8:58:19 PM2/6/14
to loj...@googlegroups.com


Le vendredi 7 février 2014 06:22:09 UTC+9, xorxes a écrit :

On Thu, Feb 6, 2014 at 1:34 AM, guskant <gusni...@gmail.com> wrote:

Although it will become out of topic, I have another suggestion related to the BPFK page of gadri.

"Any term without an explicit outer quantifier is a constant" should be changed to 
"Any term without an explicit outer quantifier can be a constant",
because an usual predicate logic has an axiom on a constant c that "F(c) {inaja} there is at least one (individual) x such that F(x)";

That applies to singular constants, whereas unquantified terms need not be singular, but the version with plural quantifiers will still be valid.


Actually, there is no explicit plural qiantifier in Lojban, though implicitly there are.

Even Thomas McKay does not adopt plural constant. For individual constant c, there are two axioms:
- [for all Y: Y {me} c] c {me} Y ;
- F(c) {inaja} there is X such that F(X) .
Even in the plural logic, F(c) implies a quantifier.

If you use the term "constant" as of the version with plural quantifiers, you should mention it in the gadri page, and also you should explain how Lojban treats plural quantifiers. Otherwise I don't understand how a constant implies no implicit quantifier.

 
 
this means that the sentence "any term without an explicit outer quantifier is a constant" automatically implicates an outer quantifier {su'o},

It shouldn't implicate that. "F{c} -> Ex F(x)" does not mean that "F(c)" and "Ex F(x)" have the same meaning, nor that "c" is just a shorthand for "Ex ...x...". Similarly xorlo says that "lo broda" is not just shorthand for "su'o lo broda".
 


I did not mean that "F(c)" and "Ex F(x)" have the same meaning, nor that "c" is just a shorthand for "Ex ...x...".
When F(c) is said, it says implicitly that "Ex F(x)" is true.


 
and it contradicts to xorlo itself that there are no default quantifiers.

Not just no default quantifiers. No implicit hidden quantifiers at all, The point is that "lo broda" is not a quantification of the bridi it appears in, the way "su'o lo broda" is.



I agree to that point, and I consider that F(c) implies implicit hidden quantifiers, and conclude that it contradicts xorlo.

 

guskant

unread,
Feb 6, 2014, 9:38:10 PM2/6/14
to loj...@googlegroups.com


Le vendredi 7 février 2014 10:54:57 UTC+9, guskant a écrit :
 
For this {lo prenu}, {by me lo prenu} is true, but {lo prenu me by} is false, so this {lo prenu} is not individual.

Sorry, I forgot {cu} :
"For this {lo prenu}, {by me lo prenu} is true, but {lo prenu cu me by} is false, so this {lo prenu} is not individual."

selpa'i

unread,
Feb 7, 2014, 6:56:57 AM2/7/14
to loj...@googlegroups.com
la .guskant. cu cusku di'e
> The way I see it, any {lo broda} is an individual (or an
> individual-collection). It doesn't matter what {broda} is. What kind of
> individuals there are in {lo broda} depends on {broda}, but they are
> still always individuals. There is no difference between {lo ckafi} and
> {lo prenu} in terms of individualness.
>
>
>
> Do you still mean
> "SUMTI is individual" =ca'e {RO DA poi ke'a me SUMTI zo'u SUMTI me DA}
> with the term "individual"?
>
> If so, keeping {lo broda} to be individual requires attentiveness on the
> universe of discourse, and reduces the flexibility of the language.

Note that I said "is an individual or an individual-collection". That
is, {lo broda} can refer to one individual or to multiple individuals,
but we are always dealing in terms of individuals. It doesn't mean that
{lo broda} must be singular, it only means that whether or not it is
plural, the only referents it has are individuals.

> Let me give an example.
>
> lo prenu cu jmaji gi'e jukpa gi'e citka
>
> I want to mean with this sentence that this {lo prenu} consists of at
> least two persons {by} and {cy}, and satisfies {jmaji} collectively {je}
> non-distributively, {jukpa} collectively {ja} distributively, {citka}
> non-collectively {je} distributively. This {lo} cannot be replaced by
> {loi} because I want it to satisfy a selbri non-collectively.
>
> For this {lo prenu}, {by me lo prenu} is true, but {lo prenu me by} is
> false, so this {lo prenu} is not individual.

Correct, because this {lo prenu} has two referents, both of which are
individuals. {lo prenu} itself is not an individual, but its referents
are individuals.

Your example sentence is perfectly fine.

Jorge Llambías

unread,
Feb 7, 2014, 3:54:27 PM2/7/14
to loj...@googlegroups.com
On Thu, Feb 6, 2014 at 10:58 PM, guskant <gusni...@gmail.com> wrote:

If you use the term "constant" as of the version with plural quantifiers, you should mention it in the gadri page, and also you should explain how Lojban treats plural quantifiers. Otherwise I don't understand how a constant implies no implicit quantifier.


There's a pretty long explanation of what I meant by constant there already, I think it's clear that a plural constant is meant:
  • Any term without an explicit outer quantifier is a constant, i.e. not a quantified term. This means that it refers to one or more individuals, and changing the order in which the constant term appears with respect to a negation or with respect to a quantified term will not change the meaning of the sentence. A constant is something that always keeps the same referent or referents. For example {lo broda} always refers to brodas. 

As for plural quantifiers, I once proposed "su'oi", "ro'oi", "no'oi" and "me'oi". 


this means that the sentence "any term without an explicit outer quantifier is a constant" automatically implicates an outer quantifier {su'o},

It shouldn't implicate that. "F{c} -> Ex F(x)" does not mean that "F(c)" and "Ex F(x)" have the same meaning, nor that "c" is just a shorthand for "Ex ...x...". Similarly xorlo says that "lo broda" is not just shorthand for "su'o lo broda".
 
I did not mean that "F(c)" and "Ex F(x)" have the same meaning, nor that "c" is just a shorthand for "Ex ...x...".
When F(c) is said, it says implicitly that "Ex F(x)" is true.

If c is singular, yes. That's not what I mean by implicit hidden quantifier though. All I mean is that saying "lo broda" is not just another way of saying "su'o lo broda" nor "[some quantifier] lo broda". 

 
and it contradicts to xorlo itself that there are no default quantifiers.

Not just no default quantifiers. No implicit hidden quantifiers at all, The point is that "lo broda" is not a quantification of the bridi it appears in, the way "su'o lo broda" is.


I agree to that point, and I consider that F(c) implies implicit hidden quantifiers, and conclude that it contradicts xorlo.

Sorry, I don't understand what you mean by that. 

guskant

unread,
Feb 7, 2014, 7:00:59 PM2/7/14