On Sun, 30 Sep 2001 ma...@kli.org wrote:
> --- In lojban@y..., "Bob LeChevalier (lojbab)" <lojbab@l...> wrote:
> > At 01:39 AM 9/30/01 +0100, And Rosta wrote:
> > >The possibility of goi-less ko'a is involved in only one of three
> key
...
>
> Maybe you guys would like to see this, which I posted on the Wiki
> s.v. "goi'a" (http://nuzban.wiw.org/wiki/index.php?goi%27a):
...
--
It's said that Mullah Omar has met two non-Muslims in his life. Others say
even that's not true.
Sami ul-Haq, Osama bin Laden's closest friend in Pakistan, runs the
"University for the Education of Truth," a fundamentalist institution that
educated and trained nine out of the Taliban's top 10 leaders.
Well, if we're not concerned with appeasing the parser, then it
really hardly matters that da'o isn't in UI. The only difference in
grammar (according to the EBNF, which I know is not canonical) is
that you can't say {da'onai}, which could conceivably be a useful
thing to say, but not the most important part of this. I think
{da'oru'e} is too wordy for single-cancel, which I think would be
very common (indeed, getting into the habit of saying {rodada'o} for
"everything" (instead of just {roda}) wouldn't be such a bad idea,
as {da} gets bound sometimes. It's wordier than it should be, but
better than nothing). Since {da'oda'o} is legal and unmistakable
for anything else, I'd say just {da'o} should be enough for
single-cancel, without being explicitly weakened. This does break
the Book, but given how small a change it is conceptually (and da'o
has seen hardly any use anyway), I somehow wouldn't feel too funny
letting usage decide this one a little differently.
~mark
.i la djan. goi ko'a klama le zarci .i la fred. goi ko'a viska lenu go'i
Are they the same person?
On the wiki, I proposed an experimental cmavo goi'a, where they are asserted
to be the same, but where
.i la djan. goi ko'a klama le zarci .i la fred. goi'a ko'a viska lenu go'i
reassigns ko'a, so they are not said to be the same. Assymetric goi does
this also, but if I understand the Book right (a big if) it is forbidden
under the baseline.
--la kreig.daniyl.
'segu le bavli temci gi mi'o renvi lo purci
.i ga le fonxa janbe gi du mi'
-la djimis.BYFet
xy.sy. gubmau ckiku nacycme: 0x5C3A1E74
The possibility of goi-less ko'a is involved in only one of three key
arguments for asymmetric goi. The other arguments are (i) that no'u
serves the function of woldemarian symmetric goi, and (ii) that ko'a
may already have a referent, which you want to assign to a cmene, and
this must be distinguishable from assigning the referent of a cmene
to a recycled ko'a.
--And.
>ko'a goi la djan. .i li'o .i la fred. goi ko'a
>You wish to interpret this as "John = Fred", instead of a reassignment?
>bi'u In usage we've been interpreting it as reassignment.
Where? "Syntactically, ``goi la .alis.'' is a relative phrase (relative
phrases are explained in Chapter 8). Semantically, it says that ``ko'a'' and
``la .alis.'' refer to the same thing, and furthermore that this is true
because ``ko'a'' is being defined as meaning ``la .alis.''." That comes out
of the refgram, chapter 7, section 5. If we define ko'a to mean Fred, and we
define ko'a to mean John also, then we could be saying that the two
definitions are interchangable (so John = Fred), or, less usefully (and
therefore not the interpretation I would advocate) that ko'a can mean either
John or Fred, and you have to glork which one it means.
>no'u probably works like you think goi already does:
>ko'a goi la djan. .i li'o .i la fred. no'u ko'a
>ko'a is John. Fred is John.
Yes, that is how goi works.
>> #(Incidentally, I don't need 2 copies of each e-mail - just reply to the
list.)
>>
>> Everyone is entitled to a vice. This is John's.
>I'm afraid John's vice is that he likes to keep the mailing list
>configured so that such duplicate mails are the default, requiring extra
>steps to overcome.
Consider the alternative. Suppose John made ''reply'' go to the list; it can
do a lot more damage to send persoanl comments to a couple hundred people
than for one person to get two copies of a mail - in which case the can
delete one. However, I do agree that people should only reply to the list,
as I am doing now.
lojbab
--
lojbab loj...@lojban.org
Bob LeChevalier, President, The Logical Language Group, Inc.
2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273
Artificial language Loglan/Lojban: http://www.lojban.org
But you seem to be failing to recognize that ko'a, like cmene,
do not have to be defined by goi. They can get their referent
from context.
English: It is big.
Lojban: ko'a barda
In neither case must it or ko'a have an antecedent in previous
utterances.
Furthermore, the referent need not even be accessible to the
addressee:
English: I read a certain book.
Lojban: mi tcidu ko'a noi ke'a cukta
ko'a ge cukta gi se tcidu mi
--And.
> At 10:54 PM 9/28/01 +0100, And Rosta wrote:
> >John:
> > > Invent Yourself wrote:
> > > > no'u probably works like you think goi already does:
> > > >
> > > > ko'a goi la djan. .i li'o .i la fred. no'u ko'a
> > > > ko'a is John. Fred is John.
> > >
> > > By no means: saying "ko'a no'u la djan." is bogus if ko'a is not
> > > *already* defined. The whole point of goi (and cei) is their
> > > defining nature.
> >
> >But you seem to be failing to recognize that ko'a, like cmene,
> >do not have to be defined by goi. They can get their referent
> >from context.
> >
> >English: It is big.
> >Lojban: ko'a barda
> >
> >In neither case must it or ko'a have an antecedent in previous
> >utterances.
>
> This may be the case in present Lojban, but when ko'a and goi were defined,
> ko'a was expected to always be explicitly defined with goi at first
> use. Other usage was *allowed* but at the risk of confusion if nonsense
> resulted.
The possibility of goi-less ko'a should not be allowed to crowbar apart
the simple solution that goi should be taken to define ko'a, and no'u be
taken to refer to an existing ko'a without reassigning it. It is enough to
establish a fail-safe usage that has solid meaning; it is not necessary to
make sure each cmavo can never be used in a dumb manner. That's where this
discussion is going. If somebody wants to intentionally be confusing and
use undefined ko'a, no'u that they feel reassign ko'a, goi that they think
don't reassign existing ko'a, or invisible ink on the back of a
watermelon, that's their option.
--
> Invent Yourself wrote:
>
> > no'u probably works like you think goi already does:
> >
> > ko'a goi la djan. .i li'o .i la fred. no'u ko'a
> > ko'a is John. Fred is John.
>
>
> By no means: saying "ko'a no'u la djan." is bogus if ko'a is not
> *already* defined. The whole point of goi (and cei) is their
> defining nature.
Yes, we agree.
> > I'm afraid John's vice is that he likes to keep the mailing list
> > configured so that such duplicate mails are the default, requiring extra
> > steps to overcome.
>
>
> http://www.unicom.com/pw/reply-to-harmful.html states my position.
That page contains fallacious arguments.
> I use two different mail clients, one of which makes it very difficult
> to remove the author as a direct recipient. Hey, I get almost all
> emails twice and some three times from certain mailing lists, and
> I've learned to live with it.
I don't want to live with it. I consider this a form of spam!
> ko'a goi la djan. .i li'o .i la fred. goi ko'a
>
> You wish to interpret this as "John = Fred", instead of a reassignment?
> bi'u In usage we've been interpreting it as reassignment.
Hmm, no, of course it's a reassignment. I must furiously to think.
> no'u probably works like you think goi already does:
>
> ko'a goi la djan. .i li'o .i la fred. no'u ko'a
> ko'a is John. Fred is John.
By no means: saying "ko'a no'u la djan." is bogus if ko'a is not
*already* defined. The whole point of goi (and cei) is their
defining nature.
> I'm afraid John's vice is that he likes to keep the mailing list
> configured so that such duplicate mails are the default, requiring extra
> steps to overcome.
http://www.unicom.com/pw/reply-to-harmful.html states my position.
I use two different mail clients, one of which makes it very difficult
to remove the author as a direct recipient. Hey, I get almost all
emails twice and some three times from certain mailing lists, and
I've learned to live with it.
--
Not to perambulate || John Cowan <jco...@reutershealth.com>
the corridors || http://www.reutershealth.com
during the hours of repose || http://www.ccil.org/~cowan
in the boots of ascension. \\ Sign in Austrian ski-resort hotel
> http://www.unicom.com/pw/reply-to-harmful.html states my position.
> I use two different mail clients, one of which makes it very difficult
> to remove the author as a direct recipient. Hey, I get almost all
> emails twice and some three times from certain mailing lists, and
> I've learned to live with it.
One of this guy's reasons (which he repeats a lot, in different forms) is that
if everyone used Elm, leaving reply-to alone would be better in every way.
Mutt handles replying to the sender instead of the reply-to just fine, but you
don't see me making decrees about how the Internet should work based on that.
His "principle of least surprise" is just shooting himself in the foot, now
that every mailing list I've seen but this one munges the header, and on this
list I often see people accidentally replying personally when they don't mean
to.
He never takes into account that it annoys people to get two responses,
especially if they sort mailing list messages into separate folders, and thus
you have to change the recipient _anyway_ to reply to a group with "Reply to
All" or people complain. As I just have.
--
la rab.spir
noi sarji le se spuda nu galfi
da'o wipes all variables, so we need a way to reassign just one, right?
or, less usefully (and
> therefore not the interpretation I would advocate) that ko'a can mean either
> John or Fred, and you have to glork which one it means.
What a horrid alternative! No way!
> >no'u probably works like you think goi already does:
>
> >ko'a goi la djan. .i li'o .i la fred. no'u ko'a
> >ko'a is John. Fred is John.
>
>
> Yes, that is how goi works.
>
> >> #(Incidentally, I don't need 2 copies of each e-mail - just reply to the
> list.)
> >>
> >> Everyone is entitled to a vice. This is John's.
>
>
> >I'm afraid John's vice is that he likes to keep the mailing list
> >configured so that such duplicate mails are the default, requiring extra
> >steps to overcome.
>
> Consider the alternative. Suppose John made ''reply'' go to the list; it can
> do a lot more damage to send persoanl comments to a couple hundred people
> than for one person to get two copies of a mail - in which case the can
> delete one. However, I do agree that people should only reply to the list,
> as I am doing now.
How often are private replies sent? And if you are sending private
material it behooves you, the one at risk of embarrassment, to protect
yourself by double-checking the mail address.
> That page contains fallacious arguments.
The burden of going forward with the argument is on you, I fear.
How are they fallacious?
> I don't want to live with it. I consider this a form of spam!
*shrug*
Killfile me if you want.
Note that you will get only one copy of this.
This may be the case in present Lojban, but when ko'a and goi were defined,
ko'a was expected to always be explicitly defined with goi at first
use. Other usage was *allowed* but at the risk of confusion if nonsense
resulted.
lojbab
Maybe you guys would like to see this, which I posted on the Wiki
s.v. "goi'a" (http://nuzban.wiw.org/wiki/index.php?goi%27a):
What would be handier (and possibly already exists) would be a
mechanism to unbind a single bound variable. I say this because
reassigning ko'as isn't the only reason you'd want to be able to do
this (besides, it isn't like there are so few ko'as that you should
ever need to rebind them much, [if you use lerfu|lerfu pro-sumti,
and why ko'a sucks].) What I see happen much more is overbinding of
{da}. People get so hung up on ''da''=something and
''roda''=everything that it's hard to remember that once you've said
''roda poi X...'' if you talk about ''da'' or ''roda'' again right
away, you're still bound in that subset of ''da'', and suddenly
''roda'' doesn't mean "everything" anymore and you have to remember
to say ''rode'' and so on. ''--mi'e mark.''
''Hence the utility of dada'o.''
Indeed. I therefore propose that ''da'o'' be used to specify
assymetry in ''goi'' and ''cei'' assignments. Whichever element is
da'o-ed is considered to be cleared out and overwritten by the new
value. This may well mean redefining ''da'o'', which I think
currently means "undefine everything." For that meaning, I propose
''da'oda'o''. DAhO has the same grammar as UI, near enough, so it
can be considered to attach to things. ''da'o'' outside of goi/cei
will retain the meaning of undefining whatever it's attached to.
This, I think, is a pretty small change, not really munging baseline
badly, and certainly it accords with grammar. And I think it neatly
solves several problems at once. ''--mi'e mark''
I second. DAhO is another example of a selma'o that should not
exist. Apparently the only difference with UI is that ''da'onai'' is
not allowed, but it has a very useful meaning: when you want to
emphasize that you are __not__ undefining something. So, whenever it
is pertinent, ''da'o'' should be moved to UI. --mi'e [xorxes]
(end of quoting)
What think you, And et al?
~mark
I think i) is dealt with by recognizing that using goi is a kind of
metalinguistic speech act. One is not merely casually noting what ko'a
means; one is actively defining it (implicit ca'e?) and more or less
insisting it be used with that definition; ii) is not something we designed
ko'a to be used for, if I understand what you mean by it. If I wanted to
assign a cmene, I would not use goi, but rather ko'a noi se cmene [name]
(or ne me'e [name])
I think that it allows us to say what we want unambiguously even with
symmetrical/woldemarian goi. So it is a satisfactory solution.
It's mildly distressing that the solutions we tend to find for our how-to-say
problems tend to involve the addition of ever more syllables and cmavo,
but I suppose I should trust in divine providence to deliver unto us
great scads of zipfings when the time is right.
--And.
> On Mon, Oct 01, 2001 at 03:49:33AM -0000, ma...@kli.org wrote:
> >(indeed, getting into the habit of saying {rodada'o} for
> > "everything" (instead of just {roda}) wouldn't be such a bad idea,
> > as {da} gets bound sometimes. It's wordier than it should be, but
> > better than nothing).
>
> Why use {da} if you're just going to cancel it afterward, anyway?
Because you want to claim existence for only a single sumti? I used this
in my rant concerning "if, then" on the Wiki, and Lojbab was confused. I
do it (use dada'o) to be tidy with my memory. Lojban contains a virtual
machine and it's a good habit to deallocate memory when you are done using
it.
The fact that
> 'roda' and 'noda' create assignments which go unused (when used to mean
> 'everything' and 'nothing') has been bothering me for a while. Why are those
> concepts not expressed by rozo'e/nozo'e, or maybe rozu'i/nozu'i?
rozo'e? rozu'i? I don't understand any of them!
certainly I agree that goi *assigns* reference.
#ii) is not something we designed
#ko'a to be used for, if I understand what you mean by it. If I wanted to
#assign a cmene, I would not use goi, but rather ko'a noi se cmene [name]
#(or ne me'e [name])
I think in this respect you're atypical of lojbanists. My sense is that there's
nothing unorthodoz about using goi to assign a reference to a name, e.g.
"le nanmu goi la djoblogz".
And whether or not I'm right about that, there is a distinction between
*claiming* (i.e. truly or falsely), using noi and ne, that X is called Y, and
*assigning* referent X to name Y (i.e. performatively decreeing that
X is called Y)..
I would have thought that computer programmers would be familiar with
this distinction (or are my memories of c. 1980 Basic *totally* obsolete?).
--And.
I haven't ever seen that, but then I have to admit I read little Lojban
text these days.
>And whether or not I'm right about that, there is a distinction between
>*claiming* (i.e. truly or falsely), using noi and ne, that X is called Y, and
>*assigning* referent X to name Y (i.e. performatively decreeing that
>X is called Y).
There is a distinction, BUT wrt names, the fact that the place structure of
cmene includes a place for those who use the name for the referent somewhat
obviates the difference. The difference between assigning a name to
something and truthfully claiming that I am using a name for something is
less than clear (obviously a false claim would not, but if one would
falsely claim to be using a particular name for something, one might well
assign a name that one intends to use for some other referent).
> At 02:14 PM 10/1/01 +0100, And Rosta wrote:
> >#ii) is not something we designed
> >#ko'a to be used for, if I understand what you mean by it. If I wanted to
> >#assign a cmene, I would not use goi, but rather ko'a noi se cmene [name]
> >#(or ne me'e [name])
> >
> >I think in this respect you're atypical of lojbanists. My sense is that
> >there's
> >nothing unorthodoz about using goi to assign a reference to a name, e.g.
> >"le nanmu goi la djoblogz".
>
> I haven't ever seen that, but then I have to admit I read little Lojban
> text these days.
I've never seen it either.
So, in {la djan. goi ko'a li'o la bab. goi fo'e li'o fo'e goi ko'a}, {ko'a}
now refers to Bob, rather than John and Bob being the same.
As to {da'o} deassigning {da}, I think it's a good idea. When Mark read my
post about the plane attacks, in which I reused {da}, he thought it sounded
like the plane was a hospital in downtown New York.
phma
Why use {da} if you're just going to cancel it afterward, anyway? The fact that
'roda' and 'noda' create assignments which go unused (when used to mean
'everything' and 'nothing') has been bothering me for a while. Why are those
concepts not expressed by rozo'e/nozo'e, or maybe rozu'i/nozu'i?
--
la rab.spir
noi sarji zu'i