>
> --- Robin Lee Powell <rlpo...@digitalkingdom.org> wrote:
>> On Tue, Apr 06, 2004 at 02:28:03PM -0700, Jorge Llamb?as wrote:
>>> {zei} makes lujvo, not the same thing.
>>
>> \/\/hatever. la mikce me la jivagos me'u, then. What, this comes up
>> in
>> everyday speech for you?
>
> (You usually won't need {me'u} there.) Modification of names with
> brivla I would class as everyday speech, yes.
>
>>> But the main advantage is the reduction in useless proliferation of
>>> selma'o.
>>
>> Not useless; it's already been shown that the proliferation allows
>> eliding of a syllable in a very, very common case, which seems
>> important
>> to me.
>
> To me it is better if you just have to learn one pattern
> (LE BRIVLA CU BRIVLA) instead of (LE/LA BRIVLA CU BRIVLA)
> and (LA CMENE BRIVLA), even at the expense of that {cu},
> but that's just me. I suspect that the possibility of
> {LA CMENE BRIVLA} existing is part of what makes the
> common error {LE BRIVLA BRIVLA} so frequent.
>
> mu'o mi'e xorxes
For what it's worth, I agree 100% with Jorge on these points.
Loglan did put la and le into the same selmaho, though despite
making the same argument about the common error le brivla brivla
(in Loglan as in lojban), I (and several others) could not convince
JCB that the cu/gu should be mandatory. If the mandatory pause
after la Alis is written or spoken, (a comma in Loglan) our parser
treats it as a gu/cu equivalent,
mu'o mi'e bab
>
>
> __________________________________
> Do you Yahoo!?
> Yahoo! Small Business $15K Web Design Giveaway
> http://promotions.yahoo.com/design_giveaway/
>
>
mi viska le sa .i mi cusku zo .djan.
is given as an example in the book, yet prima facia it's wrong: there is
no other sentence starter to erase back to.
I don't really think the incorrectness of that text is arguable
(although I'd love to see someone argue it). What I'm wondering is do
people think that allowing "sa I" or "sa NIhO" at the beginning of text
is a reasonable extension? It would be clearly labelled as
non-standard, at least in the documentation.
-Robin
--
Me: http://www.digitalkingdom.org/~rlpowell/ *** I'm a *male* Robin.
"Constant neocortex override is the only thing that stops us all
from running out and eating all the cookies." -- Eliezer Yudkowsky
http://www.lojban.org/ *** .i cimo'o prali .ui
(You usually won't need {me'u} there.) Modification of names with
brivla I would class as everyday speech, yes.
> > But the main advantage is the reduction in useless proliferation of
> > selma'o.
>
> Not useless; it's already been shown that the proliferation allows
> eliding of a syllable in a very, very common case, which seems important
> to me.
To me it is better if you just have to learn one pattern
(LE BRIVLA CU BRIVLA) instead of (LE/LA BRIVLA CU BRIVLA)
and (LA CMENE BRIVLA), even at the expense of that {cu},
but that's just me. I suspect that the possibility of
{LA CMENE BRIVLA} existing is part of what makes the
common error {LE BRIVLA BRIVLA} so frequent.
mu'o mi'e xorxes
__________________________________
But saving a syllable here and there is not that much power. The
ability to use names as brivla gives you a lot more expressive power:
ro djan poi mi djuno cu xabju le merko
Every John I know lives in the US.
> Let's leave "ease of learning" goal for the billion wanna-be
> interlangs out there, no?
It's not the main virtue of Lojban, I agree, but why make it
difficult for no good reason?
Good point. Thanks.
What exactly is the rule for sa? Can't it be "erase back until
you find a word of the same selamho as the word following sa, or
to the beginning of text if no such word is found"?
> I don't really think the incorrectness of that text is arguable
> (although I'd love to see someone argue it). What I'm wondering is do
> people think that allowing "sa I" or "sa NIhO" at the beginning of text
> is a reasonable extension? It would be clearly labelled as
> non-standard, at least in the documentation.
I'm not sure I understand what you mean. I don't see a problem
with sa at the beginning of text, it is pointless but harmless.
I'm sure you've chatted in Lojban more often than me, but isn't
"never" a bit exaggerated, given that we have chatted in Lojban
together more than a few times?
I'm not saying you can't get around it, but if the construction
was available, it would probably be used, unlike other
constructions that are available but never used.
\/\/hatever. la mikce me la jivagos me'u, then. What, this comes up in
everyday speech for you?
> But the main advantage is the reduction in useless proliferation of
> selma'o.
Not useless; it's already been shown that the proliferation allows
eliding of a syllable in a very, very common case, which seems important
to me.
-Robin
--
http://www.digitalkingdom.org/~rlpowell/ *** I'm a *male* Robin.
"Many philosophical problems are caused by such things as the simple
inability to shut up." -- David Stove, liberally paraphrased.
http://www.lojban.org/ *** loi pimlu na srana .i ti rocki morsi
That's the old meaning of me. The meaning given in CLL is:
x1 is one of the referents of "[the sumti]"
Currently only the stuff before the comma in your version. Your version
makes more sense than what I was trying, though, I think. I can imagine
horrible mistakes, though:
mi viska le broda sa la broda
(hint: LA != LE)
Of course, there is no interpretation of sa that allows that example to
do the thing the user would lkely want.
I'm *fairly* certain I've chatted in Lojban more often than you (as
you've never been on IRC). I do not remember ever having need of that
kind of construction.
-Robin
or just "la djan" (= ro la djan) in standard pre-BF Lojban.
> Probably the most explicit way to say "every john that I know",
> without the ambiguity in xorxes' suggested {ro la djan.}:
>
> ro cmima be la'i djan. ku poi mi sanji ke'a cu xabju le merko
You need "ro se cmene be zo djan", not "la'i djan", if anybody
called 'John' counts as a candidate referent.
--And.
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Ink Cartridges or Refill Kits for your HP, Epson, Canon or Lexmark
Printer at MyInks.com. Free s/h on orders $50 or more to the US & Canada.
http://www.c1tracking.com/l.asp?cid=5511
http://us.click.yahoo.com/mOAaAA/3exGAA/qnsNAA/GSaulB/TM
---------------------------------------------------------------------~->
To unsubscribe, send mail to lojban-un...@onelist.com
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/lojban/
<*> To unsubscribe from this group, send an email to:
lojban-un...@yahoogroups.com
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
It says {cmima}. I think you misread it...
--
Jordan DeLong
frac...@allusion.net
{zei} makes lujvo, not the same thing.
But the main advantage is the reduction in useless
proliferation of selma'o.
mu'o mi'e xorxes
It can't without invalidating existing sentences. e.g. {la .alis. cusku} would
be just a sumti if CMENE=BRIVLA and LE=LA.
phma
--
li fi'u vu'u fi'u fi'u du li pa
The above doesn't make sense to me....
But you got me thinking about {me}, and I think it is incorrect.
{me <SUMTI>} returns a predicate for `x1 is <SUMTI>-esque in aspect
x2'. What does {da me la djan.} mean? Unless the zo'e x2 is
interpreted to mean {leka du} it doesn't make any sense.
So I suppose you could use {du}? {ro du be su'o la djan.} is
everything equal to one of the johns.
Probably the most explicit way to say "every john that I know",
without the ambiguity in xorxes' suggested {ro la djan.}:
ro cmima be la'i djan. ku poi mi sanji ke'a cu xabju le merko
--
Jordan DeLong
frac...@allusion.net
Yes, you would need a {cu} (or a {ku}) there.
(Of course a lot of existing sentences are invalid due
to missing {cu}'s anyway, even in the most official texts.)
One additional advantage of the CMENE = BRIVLA move is
that it allows things like {la mikce jivagos} or
{la famti tom}.
mu'o mi'e xorxes
But isn't that kinda not the lojban way?
Lojban takes the approach that it's better to learn more if it gives
you more power. In this case, you spend a little to learn a little
more up front, and in return get to save syllables in the common
cases later.
Let's leave "ease of learning" goal for the billion wanna-be
interlangs out there, no?
--
Jordan DeLong
frac...@allusion.net
That's because there are too many selma'o. LA and LE should be
the same selmaho (and CMENE = BRIVLA). Also, if you said {zu}
and you realize too late you meant {vu}, you can't fix it
easily with {sa}.
I apologize, I had simply forgotten; my memory isn't what it should be.
Nothing personal.
Allow CMENE everywhere BRIVLA is allowed.
The meaning of {cmen.} as a selbri is the current {me la cmen.}
Both of which are easy to construct other ways; 'zei' in particular.
-Robin
Jordan:
> [I think you meant sanji; djuno can't be used like that]
> Realistically, you'd likely say something more like
> ro me la djan. ku poi mi sanji cu xabju le merko
I support CMENE=BRIVLA, and maybe the equivalence of "djan" and
"me la djan", but "ro me la djan" doesn't mean "every John (that
I know)" (in the usual interpretation of that phrase).
Rather, "djan" and "me la djan" would mean "every instance/subkind
of the Kind that I am calling 'djan'". (Of course, the Kind that
I call 'djan' may indeed be the Kind corresponding to the
set of all things that are called 'John' (or that I call
'djan').)
On the proposed equivalence of "djan" and "me la djan", this
would not in fact hold if CMENE=BRIVLA, since brivla "djan"
could have more than one sumti. Instead, la/lai/la'i would
simply become redundant.
--And.
Ummm, CMENE = BRIVLA? How does that work, exactly?
-Robin
The place structure would entirely depend on the meaning of the
cmene-cum-brivla. Since the meaning is glorked, so is the place
structure.
If the idea were silly, it wouldn't have been suggested by xorxes.
--And.
There's a question for you: what is the place structure of a
cmene-cum-brivla?
Not that I actually *care*: proposals that invalidate past usage don't
get much truck from me. The question is intended to indicate that the
idea is a bit silly.
-Robin
--
http://www.digitalkingdom.org/~rlpowell/ *** I'm a *male* Robin.
No, I read it right. English "every John" means "everyone
called 'John'", at least in the example under discussion.
But "(ro) la djan" (and mutatis mutandis "ro cmima be la'i djan")
means "every instance of [whatever it is I am referring to by
"djan"]". Of course, the speaker could intend la'i djan to
be the set of everybody called djan, but that's a rather
unlikely and genuinely malglico reading.
--And.
[I think you meant sanji; djuno can't be used like that]
Realistically, you'd likely say something more like
ro me la djan. ku poi mi sanji cu xabju le merko
The extra 3 syllables in that sentence are certainly going to be
more than paid for by all the elided {cu} in the rest of whatever
text/conversation is going on.
> > Let's leave "ease of learning" goal for the billion wanna-be
> > interlangs out there, no?
>
> It's not the main virtue of Lojban, I agree, but why make it
> difficult for no good reason?
I don't think it's difficult.
--
Jordan DeLong
frac...@allusion.net