[lojban] Re: Lexing text with {fa'o} and {zoi} in it

[Pierre Abbat <phma@webjockey.net>]

On Monday 03 February 2003 15:46, Robin Lee Powell wrote:
> On Sun, Feb 02, 2003 at 12:59:24AM -0500, Pierre Abbat wrote:
li'o
> >   E.  If the piece contains a consonant followed two letters later,
> >   not...
>
> Followed by what two letters later?

By a vowel. If a piece that got this far contains a consonant followed two 
letters later by a vowel, and there is no 'y' after that, either it contains 
what looks like a brivla, or someone erroneously pronounced a word of selma'o 
BY followed by a CVV cmavo without inserting the required pause (such as 
/kybusO'e/).

> I assume all the ellipses are resolved elsewhere?

The tarball, which includes the lexing algorithm (which I have some slight 
changes to make to) and the program, is at 
http://phma.hn.org/Language/valfendi-0.1.0.tar.gz.

> > 6. If the piece is a cmavo of selma'o ZOI and the ZO flag is clear,
> > resolve the next piece as the starting delimiter and set the ZOI flag
> > to a copy of it.
> >
> > 7. If the piece is a cmavo of selma'o FAhO and the ZO flag is clear,
> > set the FAhO flag.
> >
> > 8. If the piece is a cmavo of selma'o ZO and the ZO flag is clear, set
> > the ZO flag; otherwise clear the ZO flag.
>
> Seems like it would be simpler if Step 3 was:
>
> "If the ZO flag is set, lex the piece and clear the ZO flag", or
> something like that.
>
> Then you'd only have to say it once.

I think that wouldn't work, because step 5 doesn't always produce a fully 
resolved piece at the left; e.g. /zo.AisecUskuladjAn/ becomes /zo Aisecusku 
la djAn/ after two steps.

> > 9.  If there are any more pieces unresolved, return to step 3.
> >
> > According to 19:10, if {le'u} appears in a {zoi} quotation which is
> > inside a {lo'u} quotation, the {lo'u} quotation is prematurely
> > terminated. If the {le'u} is not surrounded by pauses, I don't see how
> > that can work. For instance:
> > /lo'ugA'ocUskuzoixy.vaiic.alE'umal`asOt.xy.axacuEroc.le'u/ The {le'u}
> > inside the {zoi} quotation is embedded in the Hebrew word
> > "wayish'alehu", which, being inside a {zoi} quotation, is not
> > attempted to lex (and if it did, it would choke on the `ayin of "mah
> > l`asot").
>
> Wow.  Yeah, there's no way that can work.
>
> > Chapter 21, step 2, says that the word following {zoi} *should* be
> > delimited with a pause, not that it *must*. If it is not delimited
> > with a pause, this can cause text to be lexed that is part of the
> > quotation. For instance: /zoiladjAn.bagram.ladjAn/ is lexed as {zoi la
> > djan. bagram. la djan.}.
>
> I'm sory; what is the delimiter in that case?  la?

Yes. It's ungrammatical, but the lexer doesn't know that.

> > Conversely, two words could be taken as the delimiter, as in
> > /zoi.flAlukAvbu.../ that I mentioned before. Should the word be
> > changed to "must"?
>
> Seems like it to me.

Likewise, must {fa'o} be followed by a pause?

phma