[lojban] Re: Orthography

[John E Clifford <lojban-out@lojban.org>]

The latest scheme for an alphabet for Lojban has
some nice features and, as realized, a few
problems.  It is simple, based on a simple scheme
(the 7-bit character patterns from watches and
clocks, which I will represent decimally using
the scheme  from the binary  <top, upper right,
upper left, middle, lower right, lower left
bottom>) .  It pairs voiced and voiceless by
mirror imaging, and marks stress by adding the
top line to the unstressed forms.  It includes
period and comma, which play phonological roles
in Lojban, as well as the apostrophe (or some
such mark) which has an even more pronounced
role.  So it does indeed fit the language.

However, these last three items are integrated
with the normal letters, rather than having
distinct forms to indicate their peculiar uses in
Lojban.  This is not necessarily a flaw, but, if
it could be managed, distinctiveness would be a
plus.  More problematic is the fact that the
given alphabet contains discontinuous letters,
which, while not a problem in print (except
perhaps for interpreting clusters), are a real
pain in script (it is nice to have a script form
here, but what is presented is not very good for
a running hand). Finally, the presented system
does not indicate a base line (the printed form
looks like it uses the bottom line, the script
the top) ; it does, however not contain any
character not tied to whatever baseline we
choose.  As a detail, two of the characters
repeat standard numerals, which is iffy ? we
probably shouldn?t do that if we can avoid it
but, on the other hand, the contexts of
occurrence will almost always be significantly
different so no actual uncertainty need arise.

If we were to take these objections seriously, we
might start the process anew (though keeping what
is unobjectionable in the present system ? and
the usual numbers).  Of the 127 possible
characters, 47 (by count, so I may have missed
some) are discontinuous.  If either the top line
(64) or bottom line (1) is the base, then there
are 13 more forms that are not based; if the
center line (8) is taken as base, then only two
forms drop out on this ground: the top and bottom
lines alone.  If the centerline is taken as base,
it seems a good idea also not to use that line
alone as a character, since it is easily confused
with mere transitions in running hand.  And
finally, as a practical matter for all forms of
writing, only one of each of the pairs of
mirrored  vertical lines, 2 and 4, 16 and 32, and
18 and 36, should be used, since without a
horizontal attachment they are hard to
distinguish.  This knocks 53 possible characters
out of consideration. Subtract further the ten
numerals and the 22 unobjectionable characters on
the list (actually, only 20, if we reserve the
numeric forms) and we have 45 forms from which to
extract 14 continuous characters.  Of these,  24
do not touch one of the horizontal lines, while
all the given ones touch all three.  This is a
minor matter and even less important for the
diacritical marks, comma and period (and perhaps
apostrophe).  But in any case, we end up with 23
unused but usable characters (some of the
categories above overlap).  

Since apostrophe and x do not really belong
together and certainly not as listed, we can
separate them and put apostrophe with comma and
period.  But comma and period are relatively rare
while apostrophe is quite common, hence
apostrophe should be simple, the upper vertical
halfline (16 or 32 indifferently), say.  The
other two should be distinctive but more complex
but not (as one possible line of chat would have
it) more than 5 lines, since that would make what
is a minor feature stand out too prominently. 
Perhaps comma should (following the suggestion
for apostrophe) be just the lower vertical half
line (2 or 4 indifferently).  Period is, however
more common than comma so should have the simpler
form.  In any case, there are all the lower (or
upper) corners (3, 10, 5, 12) , cup (7) and cap
(14) and Cs (11, 13) available ? not too
intrusive but distinctive. So, we need get the 9
remaining characters from the 23.

The new A (and hence a) needs to be a character
that exists both with and without the top line.
55, 60, and 63 are out because there capped forms
are numerals.  That leaves 58 (in inverted h), 59
(58 with a bottom line),  61 (the mirror of 59)
and 62 (H). The vowel characters are asymmetric ,
so only 59 and 61 (for A and E) will work.  
However, the mirror imaging  -- and, indeed, the
pairing of the vowels altogether ? does not make
a lot of sense.  I can imagine reasons to put i
and u together and, for the same reasons, e and
o, and a and y.  Or i and e, u and o, and a and
y.  The rationale for the present groupings
escapes me. (Note that 91 (E) and 27 (e) and  45
(a) are freed up to replace most of the items
used to deal with the A-3 problem. 

Next we need four left-right mirror pairs: for
bp, gk, zs, and nm. 62 and 126, being symmetric,
are well out of this.  So are 55 (square U), 63
(8 without the top), 118 (inverted 55) and 62
(H).  58 is out because its mirror is numeral 4,
82 is the mirror of numeral 7, 111 mirrors
numeral 6, 122 mirrors 9, and 114 and 116 are
arguably too close to numeral 7 (a half line
dropping from the end).  That leaves 59 and 61
(numeral 4 with a bottom line), 83 and 101
(square C), 90 and 108 (F), and 27 and 45 
(inverted F) just freed. The new pb pair frees up
the old p, 31 (mirror of numeral 6).  This leaves
x, which , not having a mate, need not be
asymmetric and thus could be something like 62
(H, but would this be a problem for English
speakers?) or any of the remaining forms.  There
are several other single forms remaining, in case
we want to rethink the pairings of l and r,  m
and n, which, while sensible, are based on
different principles from that for the first run
of consonants.

Now to see how to make this (or the original
suggestion) into a good running script.



To unsubscribe from this list, send mail to lojban-list-request@lojban.org
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
you're really stuck, send mail to secretary@lojban.org for help.