WikiDiscuss

I assume that the "lo'ai ... sa'ai ... le'ai" construct is meant to parse
as a free indicator, i.e. it can appear basically anywhere within a text.

In that case, it can easily be added to the grammar. It won't break an
already grammatical text, but it will also not fix an ungrammatical one.
So for example "le cakla kukte lo'ai kukte sa'ai cu kukte le'ai" will be
parsed as the sumti "le cakla kukte" followed by an indicator, not as
the bridi "le cakla cu kukte". And "le cakka cu kukte lo'ai sa'ai cakla
le'u" will remain just as ungrammatical as "le cakka cu kukte".

mu'o mi'e xorxes

Posted by clsn on Fri 09 of May, 2008 13:34 GMT posts: 84

Jorge Llambas wrote:
> I assume that the "lo'ai ... sa'ai ... le'ai" construct is meant to parse
> as a free indicator, i.e. it can appear basically anywhere within a text.
>
Oddly enough, for a change that has so much against it (new selma'o,
whole new construction in the language...) I kind of like this one, or
something like it. It's something we've all found ourselves trying to say.

It could be done with less violence to the grammar (though more reliance
on convention) by making {lo'ai} and {le'ai} belong to LOhU and LEhU
respectively, and just assigning a conventional meaning to {sa'ai}
inside the quotes. This would give it the grammatical standing of a
sumti (a quote), but it doesn't really have to be a free modifier. It
would probably be standing in a sentence by itself anyway.

(oh, and regarding how to use this construction to fix one of the
lo'ai/sa'ai/le'ai words themselves? That's easy: you don't.)
> In that case, it can easily be added to the grammar. It won't break an
> already grammatical text, but it will also not fix an ungrammatical one.
> So for example "le cakla kukte lo'ai kukte sa'ai cu kukte le'ai" will be
> parsed as the sumti "le cakla kukte" followed by an indicator, not as
> the bridi "le cakla cu kukte". And "le cakka cu kukte lo'ai sa'ai cakla
> le'u" will remain just as ungrammatical as "le cakka cu kukte".
>
I think this is mostly meant as a semantic fixer. I can't conceive of a
way to code the grammar such that it *could* fix a true grammar mistake,
short of really horrific stuff. But informally people would use it to
fix things it can't fix, which is probably okay.

mi'e mark.

On 5/9/08, Mark E. Shoulson <mark@kli.org> wrote:
>
> It could be done with less violence to the grammar (though more reliance on
> convention) by making {lo'ai} and {le'ai} belong to LOhU and LEhU
> respectively, and just assigning a conventional meaning to {sa'ai} inside
> the quotes. This would give it the grammatical standing of a sumti (a
> quote), but it doesn't really have to be a free modifier. It would probably
> be standing in a sentence by itself anyway.

If doing it that way, we probably can use lo'u and le'u themselves,
with just the conventional meaning that when there is a sa'ai inside
it separates the words to be replaced from the replacements. And then,
we might even just use sa for that, so that there's no need to touch the
grammar: ".i lo'u cakla sa ckakla le'u"

If lo'ai and le'ai are introduced, I think it makes more sense to have
the construct as free modifier because that's where all the meta
stuff is.

> (oh, and regarding how to use this construction to fix one of the
> lo'ai/sa'ai/le'ai words themselves? That's easy: you don't.)

Right.

> I think this is mostly meant as a semantic fixer. I can't conceive of a way
> to code the grammar such that it *could* fix a true grammar mistake, short
> of really horrific stuff. But informally people would use it to fix things
> it can't fix, which is probably okay.

Yes, but if it's informal, why not stick with s/cakla/ckakla?

mu'o mi'e xorxes

Posted by bancus on Fri 09 of May, 2008 18:34 GMT posts: 52

On Fri, May 9, 2008 at 7:04 AM, Jorge Llambías <jjllambias@gmail.com> wrote:

>
> Yes, but if it's informal, why not stick with s/cakla/ckakla?
>

Is there a compelling reason why we can't just say "skudji zo cakla na.e zo
ckakla"?

mu'omi'e.bancus.

On 5/9/08, Theodore Reed <ted.reed@gmail.com> wrote:
> On Fri, May 9, 2008 at 7:04 AM, Jorge Llambías <jjllambias@gmail.com> wrote:
> >
> > Yes, but if it's informal, why not stick with s/cakla/ckakla?
>
> Is there a compelling reason why we can't just say "skudji zo cakla na.e zo
> ckakla"?

It's too long and cumbersome for the contexts where I take this
is meant for (i.e. chatrooms). Even longer if more than a single
word is involved and you can't use "zo".

mu'o mi'e xorxes

Posted by daniel on Sun 11 of May, 2008 06:05 GMT posts: 1

> > I think this is mostly meant as a semantic fixer. I can't conceive of a way
> > to code the grammar such that it *could* fix a true grammar mistake, short
> > of really horrific stuff. But informally people would use it to fix things
> > it can't fix, which is probably okay.
>
> Yes, but if it's informal, why not stick with s/cakla/ckakla?

Because 1) you can't pronounce that, and 2) it's not as flexible (there's no convention for saying s/cakla/.../ or s/.../ckakla/ or s/.../.../) and, finally, 3) it doesn't resemble Lojban text in any way.

Besides, I don't think it has to be informal; it just can't work on the level of syntax like {si} and {sa} and {su} do. That doesn't mean {le'ai} is not machine-parsable. It just means a machine has to make the replacement and then reparse when it encounters a {le'ai} construction.

mu'o mi'e .daniel.

On Sun, May 11, 2008 at 3:05 AM, daniel <wikidiscuss@lojban.org> wrote:
> >
> > Yes, but if it's informal, why not stick with s/cakla/ckakla?
>
> Because 1) you can't pronounce that,

Is it meant to be used other than for written text? If so, I don't think it's
practicable, people don't keep a record of previous utterances in their
heads so as to make replacements in them and reprocess.

> and 2) it's not as flexible (there's no convention for saying
> s/cakla/.../ or s/.../ckakla/ or s/.../.../)

That one you are showing seems to work.

> and, finally, 3) it doesn't resemble Lojban text in any way.

Neither does it resemble English text, and yet it is used by people
chatting in English. It's not meant to be a part of the language, just
a means of presenting the language in a certain context.

> Besides, I don't think it has to be informal; it just can't work on
> the level of syntax like {si} and {sa} and {su} do.

I think that's all that was meant by "informal": not tractable by the
formal grammar.

> That doesn't mean {le'ai} is not machine-parsable. It just means
> a machine has to make the replacement and then reparse when
> it encounters a {le'ai} construction.

The way the current PEG grammar works, there is no pre-parsing
stage, which I think is a good thing. This would require reintroducing
pre-parsing. Unless it's just a free modifier, which would be fine.

mu'o mi'e xorxes

Posted by kristoun on Sun 01 of Jun, 2008 20:13 GMT posts: 2

> Use this thread to discuss the page:: le'ai

> > > Yes, but if it's informal, why not stick with s/cakla/ckakla?
> > Because 1) you can't pronounce that,
> Is it meant to be used other than for written text? If so, I don't think it's
> practicable, people don't keep a record of previous utterances in their
> heads so as to make replacements in them and reprocess.

I personally think this is very useful for verbal communication. It is, after all, very common in English speaking (thus providing supporting evidence of usefulness), it is equivalent to "I meant x not y" — to correct a mistake; "I mean, x" by itself is common to say that "x" is a correction to an unspecified error, and, "er, not y..." is an indication that "y" is incorrect, but you don't know of or are not specifying a correction right now. The correction "I mean x not y" is, I believe, a counter-example to your belief that "people don't keep a record of previous utterances in their heads so as to make replacements in them and reprocess." Indeed, I think this is *exactly* what people do when this English construct is used. I am not entirely sure what you meant by that.

Additionally, I believe the only reason people write "s/y/x" instead of "Er, I meant 'x', not 'y'" is because it's shorter and easier to type. This is probably why people see it as superior to using skudji. It's simply less words, and it remains on the meta level, as s/// does. If I could use s/// in spoken comms. I would.

As to your suggestion that we simply use s/// as is, I think that it would be nice to have a construct that is speakable.

As an aside, in my time as an IRC'er and Mumbler (two whole months) I have observed that people tend to prefer s/// over skudji, or more often than not simply say what they meant to say in their native language which kind of breaks the flow. Mentioning this is not meant to be a reason to introduce this language construct, just an observation.

On Sun, Jun 1, 2008 at 5:13 PM, kristoun <wikidiscuss@lojban.org> wrote:
> The correction "I mean x not y" is, I believe, a counter-example to your belief
> that "people don't keep a record of previous utterances in their heads so as
> to make replacements in them and reprocess." Indeed, I think this is
> *exactly* what people do when this English construct is used. I am not
> entirely sure what you meant by that.

I meant we don't do a search and replace and then reprocess the new
resulting text. We don't correct typos when speaking, obviously.

> As to your suggestion that we simply use s/// as is, I think that it would
> be nice to have a construct that is speakable.

Speakable and parseable, or just speakable but not part of the formal
grammar? My only concern is that it's almost impossible to incorporate
into the formal grammar so that it does any work, and if it's just meant as
an informal convention, then there are plenty of other options within what
we already have, so why add more cmavo?

Posted by kristoun on Mon 02 of Jun, 2008 04:01 GMT posts: 2

> Use this thread to discuss the page:: le'ai

> and if it's just meant as
> an informal convention, then there are plenty of other options within what
> we already have,

Well, a conversational convention, really. Like the vocatives. If you know of other ways to express this that aren't semantic "tricks" or annoyingly long, I would very much like to hear of them.

> so why add more cmavo?

I suppose the reason would be commonality. I think, for example, {mu'a} could be expressed with {mupli}, {mi'e} etc. I mean, how often do you say "my name is" or {mi'e}, really? Is {zo .kiris. cmene mi} a lot harder to say than {mi'e .kiris.} like "my name is Kiris" vs "I'm kiris"? For the frequency of use I doubt the value of {mi'e}. {mu'a} is probably used less than self-corrections/meta-level acknowledgement of mistakes. Indeed, I think the reason {mi'e} exists is just that expresses it precisely. Or am I wrong in that presumption? Regardless of {mi'e}'s and {mu'a}'s real reason for being included, I think there are probably more kind of meta-level cmavo that could be expressed in a couple more words in Lojban. But, as you say, if there are other ways to express this then obviously inventing more cmavo are excessive.

On 6/2/08, kristoun <wikidiscuss@lojban.org> wrote:
>
> If you know
> of other ways to express this that aren't semantic "tricks" or annoyingly
> long, I would very much like to hear of them.

One method used sometimes is to repeat the word to be corrected,
followed by si and the replacing word. That only works for single words,
but that's the most frequent type of correction anyway. For replacing
more than one word you would use multiple {si}s or a {sa}.

That of course won't make the preceding text grammatical if it wasn't
grammatical to begin with, but neither will the {le'ai} method.

> > so why add more cmavo?
>
> I suppose the reason would be commonality. I think, for example,
> {mu'a} could be expressed with {mupli}, {mi'e} etc. I mean, how often
> do you say "my name is" or {mi'e}, really? Is {zo .kiris. cmene mi} a
> lot harder to say than {mi'e .kiris.} like "my name is Kiris" vs "I'm kiris"?
> For the frequency of use I doubt the value of {mi'e}. {mu'a} is probably
> used less than self-corrections/meta-level acknowledgement of mistakes.

The main function of {mi'e} is to identify the speaker, not to say what
the name of the speaker is. If I'm talking to you on the phone and I say
{coi kiris mi'e xorxes}, I'm telling you who you are talking to, not what
my name is.

> Indeed, I think the reason {mi'e} exists is just that expresses it
> precisely. Or am I wrong in that presumption? Regardless of {mi'e}'s
> and {mu'a}'s real reason for being included, I think there are probably
> more kind of meta-level cmavo that could be expressed in a couple
> more words in Lojban. But, as you say, if there are other ways to
> express this then obviously inventing more cmavo are excessive.

I don't mind people using {le'ai} if they find it useful. What I would like
to know is what it's precise grammar is supposed to be. Is the
{lo'ai ... sa'ai le'ai} construct a free modifier?

mu'o mi'e xorxes