summary of a busy weeken, part 1 of 2

[Logical Language Group <cbmvax!uunet!cuvmb.cc.columbia.edu!lojbab>]

Here is the promised report on John Cowan's visit to DC the weekend of
the 18th-20th.  Attendees included Lojbab, John, Nora, Athelstan, and
Sylvia Rutiser, with pc joining in by phone a couple of times.

The original agenda included:

- 2nd review of papers on the Lojban tense system and MEX (mathematical
  expressions) system, with the intent of having them ready, if possible,
  for publication with JL16.

- Nora and John have been working on a formal statement of the Lojban
  morphology algorithm, and some issues needed resolution and decision.

- Deciding on all open cmavo questions, to allow a baseline of that list.

- Reviewing all open comments on the place structures and definitions
  of gismu

- Review of the progress in switching JL over to a subscription basis and
  the current fundraising drive

- Preliminary decisions on book publishing

- Determining a policy on efforts by Dave Cortesi, Bob Chassell, and others
  to put together Lojban reference books.

- Including John in a Lojban conversation session (he has never before
  been able to participate in one, since no one else in the NYC area seems
  to be actively studying).

John arrived late Fri. nite, and we started the weekend right by talking
till 4AM.  Athelstan arrived about 2AM to join the party.  Most of
Saturday was spent socializing and discussing business matters, and
various minor issues, and reviewing the tense paper.  Saturday night, we
again quit late, around 5AM this time, with everyone rising in time to
be fully awake for the Lojban conversation session.  That ended up
starting late, but the 5 of us participated in fairly lively discussion
for about 2 hours.  John had no real trouble following what was said,
and throwing his own comments in.  We then talked in English for about
an hour until Sylvia and Athelstan had to leave.

After dinner, we started on place structures, and kept going until 5AM
again.  We resumed around 11AM, and kept cranking till 5AM Tues.
morning.  Athelstan was there for all of Monday's discussions, Nora
lasted until 1030PM, since she had to work on Tuesday.  There was a long
conversation with pc in the afternoon to resolve issues that he needed a
voice/vote in.  Monday evening, we took a break from the x1's and x2's
of place structures to work on the x's and y's of MEX.

On Tuesday, John and I woke around 11AM, and kept talking till I dropped
him at the bus station around 1PM.  Whew!  Everything accomplished.

Now here is the summary of effects:

grammar

The grammar is of course baselined and frozen until we make updates and
republish it in the Lojban books.  Because we want the books to reflect
the grammar after the books are done, we do our writing based on that
next revision of the grammar.  There are now 5 changes approved for that
revision, all but one being extensions to the language.  There is also
one open issue to be resolved.  When we write and publish materials in
JL, they may be in accordance with the current baseline, or we may slip
some next-baseline grammar items in.  Because we want people to stick
with the current baseline, we are not going to distribute or talk much
about the next one until it is ready for adoption.  I will summarize the
changes in store so people know what is going on:
- adding GEK sentence GI observative as a permitted form
- adding capability for free modifiers in several new places
- adding selma'o ZEI to support a morphology algorithm change (see below)
- adding JEK+BO to parallel BO connective structures for other logical
  connectives
- correction of a precedence error, so that Ek+KE and GIhEk+KE bind more
  tightly than other connective structures

- the open issue is that, in the current grammar, it is impossible to
use a PA+MAI free-modifier after a number even though it is
grammatically legal, since the number will absorb the added PA values
(terminator BOI is not usable to solve this problem since unlike in
other situations, freemods in numbers occur before the terminator.  The
parser does not consider this an ambiguity because numbers are assembled
using the higher precedence lexer rules.  A workaround is to separate
the PA+MAI from the number with a UI word or a Y hesitation.  The
preferred solution being investigated is to add an elidable terminator
(other than BOI) to terminate subscripts, which should then allow
freemods to be attached after the BOI, instead of before.  (If this
doesn't make much sense, just consider it an artifact of the grammar
that we are polishing out and don't worry about it for now.)  The cmavo
"ja'u" is being reserved for this terminator, or for use in some other
solution that arises.

morphology

John and Nora have resolved all open issues regarding the morphology
algorithm, and Nora will be writing it up, probably to be reviewed on
Lojban List before publication in the reference book.  Problems included
strings of vowels and lujvo involving le'avla.  Since the morphology is
baselined, technically any change is a baseline change, but all changes
being considered are in areas not well-defined in the existing informal
Synopsis that describes the morphology.  Highlights (again, these are
post-book baseline features.)

- Adding selma'o ZEI, with only cmavo "zei", will eliminate various
other schemes of making lujvo using le'avla, all of which involved
either tricky stress/pronunciation problems or had potential breakdowns
of a nature similar to the Tosmabru test used in regular lujvo.  The
result would have been rules so unintuitively complicated as to make
them impractical to use on-the-fly, when most such compounds will be
made.  ZEI is processed in advance of lexer rules (as is BU for lerfu
and ZO, LOhU, ZOI, SI, SA, and SU) as part of the metalinguistic
grammar.  It causes one word immediately before it and one word
immediately after it to be considered joined into a single construct
equivalent to a BRIVLA.  With the exception of some of those
metalinguistic cmavo just listed, any Lojban word can be so joined to
any other, allowing lujvo to be based on cmavo that have no rafsi, as
well as le'avla.  Many-part le'avla lujvo will have a ZEI between each
pair of terms.  Regular gismu and lujvo may also be used as terms in a
ZEI lujvo.

- cmavo space is now recognized to include certain structures with 0 or
1 consonant, followed by more than two vowels, with apostrophe used
between every pair (except when diphthongs occur).  Thus "zo'o'o'o could
be a legal cmavo (with an obvious meaning of a more intense humor???)
These will not be considered for defined use, but are added to the
experimental cmavo space.  The grammar will treat all experimental and
undefined cmavo as if they were members of UI.

- In any string of vowels involving cmavo, an apostrophe or a pause (if
2 words), and not a glide, must be used to pronounce them.  Actual usage
has been that certain UI members have not been separated from each other
and other vowels by pauses, and this was determined to be too difficult
for the resolver to handle, so it remains forbidden.  An example is
".ua.ui" which we have been pronouncing "wah,wee", but must be
pronounced as "wah.wee".  An example showing the problems that can
result is ".ui.iu", which if pronounced without a pause is
indistinguisable from ".ui,u"

- Names will be permiited to have "la", "lai", and "doi" in them WHEN
PRECEDED BY A CONSONANT.  This means that the 'd' or 'l' must be the
at-least-2nd in a consonant cluster such that the preceding letter and
the d/l form a permissible cluster, or initial at the beginning of a
word.  This means that a name "zdoil." or "jdoil" is legal, and every
consonant except another "l" is permitted before "la" and "lai".  Thus
while "*nort.kerolainas." remains illegal, it can easily be changed to
"nort.kerlainas".  This will then allow a certain erroneous comic strip
to be corrected, by naming the cat "mlat.", "*lat" remaining illegal.
It also corrects the embarassment that the other English name of the
language - "loglan."  - has been an illegal name in "lojban."

- Names are formally restricted from having impermissible medial
consonant clusters in them.  The most significant effect of this is to
require the name "*djeimz." to be changed, since "mz" is not a
permissible medial.  "djeimyz." is acceptable.

cmavo

The following cmavo changes are made.  Note that two of the cmavo, zei
and ja'u, have grammars contingent upon the next baseline.  They will be
in the next cmavo list anyway, even though the current grammar will not
handle them.

zei     ZEI     lujvo glue       joins preceding and following word into a lujvo

ne'o    VUhU    factorial        reassigned from "zei" to make room for above

bu'u    FAhA    coincident with  space/time tense equivalent of CA

be'a    FAHA    north of         berti
ne'u    FAhA    south of         snanu
du'a    FAhA    east of          stuna
vu'a    FAhA    west of          parallel with du'a
(these are added for compatibility with languages/cultures that use a
fixed reference frame for directions instead of a speaker-based one.  A
secondary if trivial advantage is that a Lojban wind-vane is more
interesting, instead of having the letters B-S-S-S for the four cardinal
points.)

voi     NOI     descriptive clause      non-veridical restrictive clause used
                                        to form complicated le-like
                                        descriptions using "ke'a"

This is in a way similar to goi/GOI, but used with clauses (bridi) on
the right.  It defines a sumti on the left as being the thing the
speaker has in mind which fills "ke'a" in the clause.  Nick asked for
this in connection with an alternate approach to sumti raising that he
prefers to "tu'a".

Example: ko'a voi lenu ke'a cisma cu pluka mi cu zutse
The it1 whose smiling pleases me sits.

to'a    BY      lower case shift        reassigned from current "voi"
                                        tordu

ma'e    BAI     of material             used to add a material to a bridi
                                        more specific than the existing seta'i
                                        marji

de'a    ZAhO    pausitive               event contour for a temporary halt
                                        and ensuing pause in a process
                                        denpa
di'a    ZAhO    resumptive              event contour for resumption of a
                                        paused process

   mi de'a citka ca lenu la noras. tavla
   I pausitively eat while Nora talks.

vu'i    LUhI    the sequence            converts other sumti types to
                                        sequences, even if order is
                                        unspecified

ja'u    JAhU    reserved                reserved for possible solution
                                        to PA+MAI problem mentioned above

remaining unassigned (22):

bi'a bi'e bi'u bu'o
bo'a bo'e bo'i bo'o bo'u
ce'e ce'u ci'a
do'i
ju'e
mi'i
na'a ne'e
re'u
te'i
va'e va'u vu'o

gismu are discussed in the second part of this message

John Cowan will hopefull discuss sets and masses and sequences in more detail
in a separate post, and will of course correct my errors in these messages
since he ahs a better memory than I do.

----
lojbab = Bob LeChevalier, President, The Logical Language Group, Inc.
         2904 Beau Lane, Fairfax VA 22031-1303 USA
         703-385-0273
         lojbab@grebyn.com