Proposed changes to the Lojban lerfu system

[John Cowan <cbmvax!uunet!marob.masa.com!cowan>]

The following is a proposal to simplify and streamline the parts of Lojban
related to lerfu (letterals).  The present system is summarized, a revised
system proposed and explained, and a rationale for the suggested changes
presented.  Information on the present system is drawn from LLG documentation,
including the >Overview of Lojban Grammar<,
as corrected by the 5 May 1990 edition of the machine grammar.


THE PRESENT SYSTEM:

At present, 42 cmavo are assigned for letterals and related purposes, grouped
into seven lexemes known as BY, NEI, FAU, VOI, NAU, FOI, and SIhE.

BY lexeme includes the 23 letterals themselves:  abu, ebu, ibu, obu, ubu
for the vowel letterals; by, cy, dy, ... zy for the consonant letterals,
and y'y for the vowel buffer letteral.

NEI lexeme signals an "alphabet selection shift":  members of NEI cause
letterals to be interpreted as representing characters from non-Lojban
alphabets.

	nei enables the Lojban alphabet;
	ra'o enables the Arabic alphabet;
	ru'o enables the Russian alphabet;
	ge'o enables the Greek alphabet;
	jo'o enables the Hebrew alphabet;
	de'a enables part 1 of the Hindi alphabet;
	de'o enables part 2 of the Hindi alphabet;
	la'o enables the Latin alphabet.

FAU lexeme signals a "special character shift": members of FAU cause 
letterals to be interpreted as representing non-standard letters or other
characters.

	fau enables up to 22 more letters from the current alphabet;
	lau enables up to 22 non-letter characters;
	ba'o enables up to 22 user-defined characters.

VOI lexeme signals an "upper/lower case shift":  members of VOI cause
letterals to be interpreted as upper or lower case.

	ga'e enables upper case;
	voi enables lower case;
	tei toggles case (enables upper if lower is enabled, or vice versa);
	tau toggles case for the next character only;
	to'o toggles case for the next word only.

NAU lexeme signals "default case shift".

	nau cancels any current FAU or VOI in effect.

FOI lexeme signals an "operator set shift".  A string of letterals may be
used within MEX to represent a mathematical operator, if preceded by the
marker ma'o.  These operators are grouped into operator sets, each of which
is numbered by a numeral (string of digits).

	foi + numeral + BOI enables the operator set specified by the numeral.

SIhe lexeme signals "default operator shift".

	si'e cancels any FOI currently in effect.

Members of NEI, FAU, VOI, NAU, and SIhe have null grammar.  Sequences of
the form FOI + numeral + BOI also effectively have null grammar.


PROPOSED REVISION:

Only 34 cmavo are used, grouped into only three lexemes:  BY, FAU, and GAhE.

Lexeme BY contains the existing 23 letterals, plus two new ones for
representing the remaining two characters of the Lojban alphabet.

	sa'a (derived from slaka) represents the close-comma written ",";
	si'e (derived from sisti) represents the pause written ".".

Lexeme FAU contains the letteral shift characters that must be followed by BY.

	fau + BY sets the letteral set;
	lau + BY sets the 2nd-order letteral set.

Lexeme GAhE contains all the letteral shift characters that have null grammar.

	ga'e enables upper case;
	voi returns to default lower case;
	tau enables upper case for the next letteral (or double letteral) only;
	nau resets the letteral set to default;
	nei resets the 2nd-order letteral set to default;
	ba'o enables double-letteral mode;
	to'o disables double-letteral mode.


DISCUSSION:

The cmavo "ga'e" and "voi" are meant primarily for spelling out
Lojban cmene, which may contain upper case letters to mark terbasna.
For example, the name ".eiRIK. TIdeman." would be spelled out as follows:
	si'e. .ibu. ga'e. ry. .ibu. ky. si'e.
	ty. .ibu. voi. dy. .ebu. my. .abu. ny. si'e.
Since pauses may be inserted between any Lojban words for added clarity,
"si'e" may be inserted between the spelled-out versions of words as well.

The cmavo "tau" is not likely to be used in a name, although the name "Rl."
(= English "Earl") could be spelled "tau. ry. ly. si'e".  Instead, it
is meant primarily as a prefix when using letterals as symbolic names,
to provide more symbolic names.  For example, in Newtonian mechanics
the symbols "g" and "G" are commonly used with different meanings;
Lojban would represent these as "gy" and "taugy" respectively.

The cmavo "fau", "lau", "nau", and "nei" give access to the full range
of representable characters.  The assignments of the 25 lerfu to the specified
uses apply only by default: they constitute the "default letteral set".
By saying "fau.abu" we remap the 25 letterals onto a new (non-Lojbanic) set
of characters.  Similarly, "fauby" makes another 25 available, and so on up
to "fausi'e", for a total of 625 new characters.  To restore the Lojbanic
interpretation, use "nau".

If 650 characters are not enough, "lau.abu" changes to a new set of 650
characters, and so on to "lausi'e".  The cmavo "nei" restores the default
set of 650.  The total number of characters available is thus 16,900, which
can be doubled by the upper/lower case shifts to a grand total of 33,800.

When using alphabets which have many more than 22 characters, constant flipping
of the fau switch may become uncomfortable or verbose.  The cmavo "ba'o"
enables a special mode of letteral interpretation called double-letteral mode.
In this mode, two letterals are needed to represent a character.  The first
one specifies the letteral set; the second, the letteral within that set.
This is equivalent to placing "fau" before every pair of letterals, but
saves space and time.  To turn off double-letteral mode, use "to'o".

The operator shifts currently specified by FOI and SIhE are no longer
handled in this part of the grammar.  Instead, they are specified by
placing subscripts on ma'o.  Thus, ma'oxipa signifies that the following
letteral operator (and any to come) are drawn from the first operator set;
ma'oxire enables the second set, and so on.

The cmavo ba'o, foi, ra'o, ru'o, ge'o, jo'o, de'a, de'o, la'o, and tei are
now freed up for other uses.  The currently unassigned cmavo sa'a is assigned.


RATIONALE:

I believe that the proposal above is both more complete and simpler than
the existing system.  Obviously, all things being equal, three lexemes are
better than seven, especially when the grammar could not distinguish between
five of the old lexemes anyway (all had null grammar).  This could be
achieved simply by lumping the old lexemes together, however.

The proposed system handles Lojban's own needs better, by providing for
exactly those constructs that are needed to write Lojban itself:  the
25 characters in the Lojban alphabet, the two cases (upper and lower) and
the space between words.  (Space and pause are not distinguished, but the
language allows every space to be replaced by pause anyway.)

The list of alphabets in old lexeme NEI is incorrigibly miscellaneous.
It neither represents all the world's alphabets nor even all the important
ones (the omission of Japanese is notable).  In particular, it has to handle
Devanagari with two cmavo; Japanese might require four just for hiragana/
katakana, and Japanese kanji/Chinese hanzi are totally beyond its reach,
even if every existing cmavo were assigned to this one function of handling
foreign alphabets!  The proposal, on the other hand, deals with over 25
cubed different characters without ambiguity and with only four cmavo.
Some conventional assignments could evolve if some alphabets are more
needed in running text (especially MEX text) than others:  for example
"fauly" might come to be used for the non-Lojban Latin letters, and
"faugy" for the Greek letters.

The treatment of operator sets using ma'o subscripts above is an "optional"
part of this proposal.  The language in the >Overview< implies that lerfu-type
MEX operators are single lerfu, whereas the grammar describes them as lerfu
strings.  With lerfu strings available, the whole idea of needing separate
sets of operators may be in fact obsolete.  If not, the subscripts on ma'o
at least put the burden where it belongs, within MEX.

Please send comments on the above to cowan@magpie.masa.com, to the lojban-list,
or to:
	John Cowan
	Chemical Bank, Cooperative Systems
	95 Wall St., 6th floor
	New York NY  10005  USA

-- 
cowan@marob.masa.com			(aka ...!hombre!marob!cowan)
			e'osai ko sarji la lojban