rafsi proposals

["Mark E. Shoulson" <cbmvax!uunet!ctr.columbia.edu!shoulson>]

Well, I suppose I have to come up with some opinion on these proposals
(sorry for my relative silence lately, I've just returned and have a whole
lot going on).

John's probably right as far as not assigning "unneeded" rafsi: better safe
than sorry.  I also think some of lojbab's proposals are needlessly
nitpicky.  Moreover, I doubt the utility of the corpus of data we have.
Lojban hasn't been used to write in for very long, and what has been
written in it has tended to be rather specific (to the interests of the
>very small< group of writers.)  What's more, nearly everything written in
it so far has been translated, and (speaking at least for myself) not
necessarily translated very well.  There seems to be an emphasis on
duplicating the usages of English (or whatever the original language is),
that only a very few have been able to start to break away from, and at
that only inconsistently (please, guys, no offense.  I'm at least as bad as
everyone else, probably worse, but I can see the problem anyway).  This
tendency hits lujvo-making particularly hard, as a writer tries to create a
*word* for what is a word in the source language, when a tanru or phrase or
totally different (dare I say "alien"?) sentence structure might be more
elegant or "Lojbanic".

I'm not sure I agree with some of lojbab's comments on the relative
preferences of different rafsi forms (and if someone brings evidence from
my own work, I'll still say so: nobody promised you consistency).  I
personally feel that CV'V rafsi are not terribly useful as final elements,
since they don't save you any syllables over the unreduced 5-rafsi, and
syllables are vastly more important than letter or sounds when it comes to
speaking.  The only advantage I can see to them as finials would be in a
case where the rafsi started with a letter that required less hyphenation
before it (for common lujvo), and that's a tough thing to analyze.  As
initials, I'd prefer CV'Vs to CVCs in cases where the final C has lousy
hyphentaion prospects.  That is, I'd rather use r/n hyphens than y hyphens
(I think this conflicts with what lojbab says).  Of course, the C would
have to be a really lousy letter, and dependent on common lujvo, so this
isn't a very objective statement.  Thus, I'd argue that a gismu with a CVC
rafsi, particularly one with a "good" final C, doesn't need a CV'V (though
a CVV might be worthwhile, if available), and gismu likely to be found only
in final position also would benefit very little from a CV'V: either give
'em CCV or CVV, or save the CV'V for someone else.

The whole idea of reassigning the rafsi on such a grand scale makes me feel
like I'm standing on a shifting sand-dune, but that's my own prejudice, and
the rafsi really do need an overhaul, so I'll try to ignore it.  Even so,
I'm less than sanguine about making so many changes, particularly those
with not-so-much justification (backward-compatibility is nice, but you
can't make an omelette...).

I don't know if I can give opinions on *every* proposal, as Nick and John
have done, but I feel like doing something long today, so I'll try.

1.  YES.  John's right about {selski} and {velski}, even though he calls
them {selsku} and {velsku}.

2.  SHRUG.  Note that my "SHRUG"s indicate more negativity than Nick's
probably do:  I'd rather not change more than necessary.  {bajra} almost
certainly doesn't {baj}, at least not so badly as to cause reassignment.

3-5.  YES.  Cultural gismu are rightly de-emphasized, and "success" is
important.

6.  SHRUG.  Actually, perhaps a tentative YES.  I suspect {cmila} has more
to it than Nick might suspect, and in non-final positions at that
(laughingly-say, laughingly-cry out, etc...).  (Recall that I see little
advantage to CV'Vs as finals).

7-11.  SHRUG.

12.  YES.  {nanca} could be productive.

13.  SHRUG.  I think I like Nick's proposal.

14.  YES.  Since {larcu} has {lar}, by me reasoning about CV'Vs, it doesn't
really need {la'u}.

15.  YES.  I prefer the proposal, weakly, to the alternative.  {ciksi} and
{cinse} and {crise} will benefit from the better rafsi they'd receieve.

16-18.  YES.  I see advantages and little in the way of disadvantages.

19.  SHRUG.  I kinda liked {gar} in {gapru}, but I wouldn't be upset to
lose it.

20.  SHRUG.

21.  Hmmm.  {mau} is a little less sacred to me than to Nick, though it's
seen a *lot* of use.  But {zma} can replace it in many places.  I'd rather
not see it go, though, and I agree with John that if it does I don't want
it in {cmalu}.  Then again (going against my usual preference to keep with
backward compatibility), why should we inflict upon posterity our archaic
preferences?  No vote from me.

22.  YESBUT.  Nick is right: {simxu} will (or should be) heavily used, in
final and non-final positions, and deserves {sim}, not the icky {six}.  I'd
support his alternate, perhaps minus the shenanigans to give {si'u} to
{simxu}.

23.  YES.  I like {nan} for {nanvi}, but {snanu} needs it more.

24.  NO.  I agree with Nick.

25.  YES, with reservations.  {tcaci} may turn out to want {tca} more than
{tcadu}.

26.  SHRUG.  Don't see the point very much (see above about CV'Vs).

27.  SHRUG.

28-31.  YES.  I kept wondering why I see in Nick's writing things like
{tcebarda}--"litre-big".  Huh??  Then I realized he probably meant
{mutcybarda}.  Much better (though I'd probably prefer {je'a barda} or
something.  {mutce} is still gobs more productive than {litce}).  And {bac}
looks to be useful too.

32.  SHRUG.  I don't see much point.

33.  Less enthused by it than lojbab.

34.  Reluctant YES.  {ka} deserves a good rafsi.

35.  SHRUG.  Doesn't hurt {morsi} much, but I agree that there's little
point to a final {moi} rafsi.  Not too sure about that, though...

36.  YES.  {ni} deserves it, and {nilce} can manage.

37.  YESBUT.  I also would rather {tol} for {to'e}, somehow it fits my mind
better.  Nick points out that {toldi} probably doesn't need the rafsi,
which is likely true, but it is also true that (in my lists anyway),
{toldi} doesn't *have* {tol}.  I think it's unassigned, which would make
everyone happier.  Giving {nor} to {no'e} (thus obviating the need to give
it to {no} and taking {non} to {no'e}) wouldn't be too horrible either, but
it's pretty much six-of-one/half-dozen-of-other, aside from inertia
considerations.

38.  YESBUT.  {romge} barely deserves the gismu, let alone two {rafsi}.

39.  YES, agreeing with John and Nick.

40-46.  YES.

47.  SHRUG.  Not positive it's justified.

48.  Much ado to get a rafsi for {gerku}, which probably doesn't need it
that much.  While {gle} would do well in {gletu}, I'm not sure it's worth
the effort.  Tentative NO.

49.  YES.

50.  SHRUG.

51.  SHRUG.  Not sure {tamji} will see that much action.

52.  SHRUG.

53-56.  YES.  {creka} probably should have a final form, but {certu} might
need it more, and maybe {cet} should be freed, since {certu} will have a
CCV.

57-58.  SHRUG.  The gain seems minimal, but there isn't even a
backward-compatibility argument against it.

59-60.  SHRUG.  Don't see quite so much advantage to 60 as others.  Both
could be productive, but this is not a very negative SHRUG.

61-63.  YES (re: #62: certainly changes translations of "of all the rotten
luck!"  zo'o).

64.  Agree with Nick.  Insufficient data.

65.  NO.  Lack of productivity.

66-67.  YES.

68.  SHRUG.  I can see some productivity from {ladru}, but not very much.

69.  SHRUG.

70-71.  YES.

72.  SHRUG/NO.  At first glance, it looks good, but {jundi} is more likely
to be non-final, and {judri} final.  reversing the assignments might be a
better plan, though it'll conflict with the cmavo {ju'i}.

73.  YES.

74-75.  SHRUG.  Can't see much advantage.  in #75, it's switching a CV'V
between two gismu that are candidates for final members, not yielding much
advantage.

76.  NO.  I also agree with John.

77.  YES.

78.  SHRUG.  Also can't see much point for {mei} getting {mei}.  John's
alternative sounds acceptable as well, and likely better than the proposal.

79-80.  YES.

81.  NO.  Nick's and John's reasoning makes sense.  Thirst is thirst.

82-83.  YES.  Not sure how helpful #83 will be, though.

84.  NO.  The lujvo made with {dei} *should* be {djedi}, not {dinri}; it's
not malglico.  I suppose I could live with replacing those with {dje}, but
I'd rather consider it sacred.  {dinri} does deserve a rafsi, though.

85-86.  YES.  Arguably, {virnu} is more likely to be initial, and maybe
could be given a CVC or something, while the more valuable CCV could be
assigned somewhere more important (not to {vidru}, it doesn't need
something as good as a CCV).

87.  SHRUG.  Much clanking and clunking for little more than better
correspondence between cmavo and rafsi assignments, which we never promised
and can't deliver in general anyway.  Leave it alone.

88.  YES.

89.  YES.  Be interesting to see how to use it...

90-92.  SHRUG.  Much ado about little.

93.  YES.

94-95.  SHRUG.  Don't see much point.

96-97.  YES.  possibly productive.

98.  Much clanking about lotsa stuff, probably most will not be used.  Even
{jve} is probably not going to see heaps of use.  Bear in mind that when
these cmavo-rafsi show up in lujvo, they'll always (except in truly bizarre
cases) be between two elements, so CVCs are fine for them (though CCV makes
for no hyphens).  NO to the lot, with a slight reservation on {jve}; who
knows, it might be worth it.

99.  Can anyone think of a reason to have anything other than a CVC for
{ce}?  It, too, will always be a middle element, and {-l} is excellent wrt
hyphenation.  Overkill.  NO.

100-104.  NO.  Maybe 101-103 have some merit, but probably not too much.
OK, SHRUG on those three...

105.  SHRUG.  Not sure.

106-134.  Most of these are probably not worth it.  Tentative NO to the
lot, modulo some exceptions.

135-136.  NO.  Tough enough to keep similar rafsi from unrelated gismu
straight, let alone similar rafsi from similar cmavo.  Better to stick with
self-rafsi and take the extra syllables; these aren't going to be too
heavily used anyway.

137.  SHRUG/YES.  Probably not all that useful, but {vraha} wasn't going to
use {va'a} all that much, and I like r/n hyphens better than y hyphens, so
it's probably for the best.

138-140.  YES.  {dzipo} is not a high-flyer gismu in the first place, and
{cfika} could use the help.

141.  SHRUG.

142-143.  YES.

144-145.  SHRUG

146.  SHRUG.  {rirxe} is likely to be final, where {ri'e} alone won't do
all that much.

147-149.  SHRUG.  Not sure if robbing {kelvo} of {kev} is a good idea,
though.

150.  NO.  I'm not so vehement as Nick, but his point is likely valid.

151-152.  YES.

153.  SHRUG.

154.  YES.  Note, though, that {girzu} appears in some of the early docs,
and its rafsi are discussed, so there may be a (very small) issue of
sacredness.

155.  YES.

156-158.  SHRUG.

159-161.  NO.

----------------

OK, there you have it.  Have a blast....

~mark