[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[no subject]



Following is the near-final report on the rafsi review, incorporating
comments from Nick, Mark, Colin, John Cowan, Nora and David Twery, with
all issues looked over at least twice.  A few changes remained undecided
because the decision needed to be based on the supporting data, which
I've added herein.  A few others were modified per your comments, and
presumably will be acceptable.  I will presume that the proposals as
explained will be approved unless people object, since except as noted,
Cowan, Nora, Twery, and I all agreed to them at LogFest.  But feel free
to speak up.  I will minimize repetition of data from the earlier report
to keep this short.

Note that I cannot easily draw in from the hundred or more pages of
handwritten notes of lujvo proposals in the files.  Some of these lujvo
came from the Roget's work, hence help give breadth of semantic
coverage; others were proposed in the pages of The Loglanist, or by
Nora, Tommy and me when we were arguing about what should and shouldn't
be in the gismu list.  There also may be redundancy, of course, since I
have no easy check to see if a proposal has been made more than once in
Loglan history.


Approved without further comment

1, 3-4, 10, 12, 14, 16-18, 23, 26, 28-31, 34, 36, 41, 44-46, 49-50,
53-54, 56-58, 61-63, 66-67, 69-71, 73-75, 77, 79-80, 82-83, 88-89, 93,
96, 101-103, 139-144, 151-153, 155-157

Approved, but comments/justification provided.  Can be reconsidered if
someone asks.

5, 8-9, 11, 19-20, 25, 32, 33, 35 (78), 40, 43, 47, 48, 51-2, 59, 64,
68, 86, 90, 97, 106, 111, 113-4, 121-2, 132-4, 138, 146, 148

Modified/Open Issues  Please Revote.  Benefit of doubt towards approval.

2, 6-7, 13, 15, 21 (76), 22, 24 (150), 27, 37 (38), 39, 42, 55, 65, 84, 85
91-92, 95, 97a, 98-99, 105, 107-8, 110, 116-7, 135-7, 147, 149, 158

Rejected/Withdrawn based on comments

72, 81, 87, 94, 100, 104, 109, 112, 115, 118-20, 123-31, 145, 154, 159-161

Alternate proposals B&C received no particular support as written.
Alternate proposal A was partly incorporated into a revised #15.  Alternate
Proposal D was rendered moot by changing the gismu.

Cowan asked me to check unmodified status quo for 'overkills' with CCV
and something else that can be deleted per the standards set in this
review.  These are at the end, numbered starting from 201.  I erred on
the side of the status quo on these.  I tended to keep 'natural' CVCs or
CVCs starting with a different letter than the CCV when there was
significant 1st position usage, and to drop CV'Vs.  Only the changes are
given unless I had a special comment.

Even with this conservative approach, you will note that several of the
cases where there is a CVC and a CCV available, people have regularly
chosen the CVC, sometimes even when hyphenation is needed, and sometimes
when the CVC isn't even the 'natural' one (e.g. jaf/jamfu and
gic/glico).  This to me calls into question the whole idea of
'overkill', and suggests that people prefer a CVC to almost anything
else if they can find one.  Accepting this as a factor SHOULD cause us
NOT to make the changes associated with overkill, and indeed to go the
other direction if possible.  But I leave the vote to you guys.  A
general vote on the question of deleting rafsi due to 'overkill' when
there is already a CCV is appreciated, and any consensus will cause me
to globally undo other changes in this report without further vote.
Again, as with the case of CV'V in final position, what logical analysis
tells us 'should' be good, does not seem to reflect what people choose
in practice.  I have been a maximalist on rafsi assignments to give
people in the post-design phase more choices rather than make the
choices in advance for them.  Who are we to say that Chinese Lojbanists
wouldn't prefer vowel rich CVVs (with or without the ') over the
consonant clusters of CCV and most CVC compounds?  On the other hand, I
am willing to accept the minimalist argument, as long as all you people
are sure that is the right thing to do.  (I can also, if people want, be
more agressively minimalist - there are another 30 or so instances of
CVC and CCV for a single word, generally each starting with a different
letter, which I omitted from changing because hyphenation may justify
keeping both.)


_____________
ITEMIZED LIST
-------------

2. NSN considered bal to be very useful on banli and I agree.  The
change that gives bra to barda, however, makes that word a bit more
effective than it used to be in initial position, weakening this
argument.  He claims bajra is only used in bajrykla, put the data shows
otherwise (and this will after all be a much used word in combination
with klama, litru, plipe, etc.)  If there is still strong feeling on
banli, the change of baj is separable from the rest, since bajra has the
statistics to take baj from balji even if balji doesn't get 'bal'.  But
then the net benefit drops to 5 (actually 11 after correcting the typos
noted in the actual data).  I would not argue with this modification,
given examination of the data, though the Eaton data, while profusive,
is of course highly suspect.

MS agreed with Nick on bajra.

baj *   *    29 13 baj          bajra            29  16   1  12   3  13 run
bal *   *     1  1 bal          balji baj        14  13   1   3   8   1 bulb
-   *   ba'i 16  0         ba'i banli bal        89  73  24  28  14  16 great
*   *   -     0 15 bav          balvi bavba'i    97  82  21  40  15  15 future
----- net benefit 17

Supporting data
Lojban usage
balji:  (all three of these appear to be typos that actually support bajra)
10   terbajli'a
1    bajli'a
8    sukybajykla

bajra:
37   bajryjvi
1    bajrykla
19   bajrystu
1    cidbajrygre
2    cidbajrykla
3    nunbajra

banli:
12   balcu'e
2    baldakfyxai
48   balkumfa
2    balrai
2    baltutra
1    balxau
36   mlebanli
1    vribanli

balvi:
2    ba'ispe
3    ba'ivla
2    lamba'i
1    bavla'i
11   bavlamdei
9    bavlamjeftu
11   bavlamji
2    bavma'a
1    bavseljbe

TLI Eaton proposals

balji - none
bajra - 5 usages in final position that using our place struture would
require bajrykla or some relative.  These were not counted, since I made
no such modifications to Eaton data based on place structure
differences.  But they further enhance the benefit to bajra.

pagre +bajra
darno +bajra
darno +bajra
darno +flalu +bajra
drata +stuzi +bajra

banli:
nu +banli +jikca +dansu
banli +snada
banli +jvinu
melbi +banli
carmi +banli
natmi +banli
banli +sanmi
banli +sinma
banli +natmi
srana +banli +se +ciska
banli +mukti +xebni
banli +sidbo +prenu
muvdu +banli +melbi
jbena +turni +banli
banli +jikca +dansu
melbi +banli +jinga
banli +turni +ninmu
melbi +banli +penmi +tavla
banli +jubme +since
banli +jbena +turni
banli +dansu +fasnu
gunka +banli +fuzme

balvi:
pa +balvi +pa
na'e +purci +na'e +balvi
balvi +cliva
balvi +jmive
balvi +krici
dunra +balvi
citri +balvi
balvi +cinse
balvi +senpi
balvi +ciska
balvi +tavla
punji +balvi +ti
srera +balvi +zbasu
sorcu +dunda +balvi
dacti +balvi +morsi +fo'a +se +ciska

TLI dictionary
balji 1 initial, 1 final
banli 7 initial, 2 final
balvi 11 initial 1 medial 6 final
bajra 1 initial, 2 final (equates to net 3 non-final due to place structures)


5. NSN questioned the evidence for this one, though he voted yes.
Rechecking the data shows some clear benefit to sanji, with little
negative to pajni since it retains its Cvv.  Presumed approved.

saj *   *    26  8 saj          sanji            26  18   0   7   5   8 consci
paj *   *     4  4 paj          spaji saj        17  13   3   5   0   4 surpri
-   *   *     0  0         pai  pajni pajpai     63  20   0  10   4  43 judge
----- net benefit 18

Lojban usage:

sanji
2    cmesanji
9    nalsanji

spaji: none

pajni
6    pajvrude
12   jvipai
3    maptypai
2    pantypai
7    selxlupai
2    xaurpai
1    zugypai

Eaton data
sanji
sanji +cirko
spaji
bebna +spaji
suksa +spaji
barda +spaji
mutce +spaji
spaji +suksa
pajni
pajni +prije
zifre +pajni
purci +pajni
casnu +pajni
zifre +zukte +jicmu +pajni
catlu +claxu +pajni
rarna +drani +pajni
galtu +pajni +stuzi
vamji +pajni

TLI dictionary
pajni 2 initial 2 medial 4 final
sanji 1 initial 1 final
spaji 2 initial


6. NSN feels that cmila needed justification.  MS saw some justification
and proposed two cmila initial position lujvo.  The supporting data thus
gives 7 initial, 2 medial, 5 final for cmila, all 14 of which are
supported by mi'a.  Stretching to include the TLI lujvo, jmina has 3
initial and 7 final, thus giving 3 supportable by min and 10 supported
by mi'a.  But I will note that the jmina lujvo have had significantly
more usage, so if word frquency is allowed for, the comparison is much
closer.  NSN's argument in favor of mid for midju isn't convincing to
me, but I will agree that the changed set of rafsi results in less
'natural' assignments than the old set, and minji is left with a fairly
mediocre rafsi for those times when it is found initially.  I won't
fight too hard for this one if Nick still thinks it is bad.

*   *   mi'a 26  0         mi'a cmila            26  17   3   5   2   9 laugh
min *   -     0 12 min          jmina mi'a       30  18   0   9   3  12 add
mid *   *     0  0 mid     mi'e minde minmi'e    81  47  12  13  15  34 comman
mij *   *    11 11 mij          midju mid        50  39   8  21   5  11 middle
-   *   *     0  0         mi'i minji mijmi'i    37  12   2   4   2  25 machin
----- net benefit 14

Lojban usage
cmila
2    cmacmila
1    cmilype'u
1    cmilyri'ape'u
1    cmilysatre

jmina
1    jonmi'a
8    mi'acru
10   mi'arbe'i
1    mi'arci'a
6    poirmi'a
7    selmi'a
1    velmi'a
1    zbaselmi'a

minji
1    mi'ispi
1    taxmi'i
2    tergu'imi'i

Eaton data
cmila
nu +cmila +culno
se +cmila
suksa +cmila
spoja +cmila
cmila +spoja
zmadu +cmila +dacti
sutra +cenba +cmila
cmila +krinu +pilno

jmina has no TLI gismu equivalent; it is a 2-place lujvo with 2 final
position equivalents in the TLI dictionary; cmila has a TLI gismu
equivalent but had no lujvo based on it in the dictionary.

minji
muvdu +zbasu +mlana +minji
minji +gidva +bloti
3 initial 6 final in TLI dictionary

7. CF had a strong dislike for this one, the only one he objected to,
asking for data on karce, kancu, and kalsa.  Mark shrugged, Cowan
thought karce initial was fairly worthless.  I think the data supports a
switch in favor of kalsa, and kancu seems quite untouchable even
questioning the TLI data.

kas *   *    25  0 kas          kalsa            25  25   1  19   0   0 chaoti
kan *   *    19 19 kan          kansa kas        79  60  16  23  13  19 with
kac *   *     6  6 kac          kancu kan        32  26   9   9   3   6 count
-   *   *     3 14 -            karce kac        21  18   0   8   3   3 car
----- net benefit 14

Lojban usage
kalsa
6    kalsyfle
11   kalsygri
4    kalsykla
3    kalsyklo
2    kalsyri'a
3    kalsysu'a
kancu
26   cmekancu
2    kangau
1    kankre
2    kanri'a
karce
8    kacplu
(and Colin's proposal kackla)

Eaton data
karce
nu +muvdu +karce +bloti
kancu
na'e +cumki +kancu
na'e +kancu +cumki
na'e +se +kancu +gasnu
na'e +se +kancu +zbasu
kancu +claxu
kancu +zbasu
kancu +gasnu

kalsa has no TLI equivalent defined, either gismu or lujvo
karce 1 initial 2 medial 3 final in TLI dictionary

TLI's kancu is not quite the same as ours, which is a 2-place lujvo
(kancu gasnu) for them.  Their kancu would be a conversion of ours:
terkancu.  The other TLI dictionary lujvo based on their kancu is
equivalent to an initial position usage of our meaning.


8. JC found both of the following rather useless in lujvo, but all voted
yes.  Here is the supporting data.  The D&D terminology obviously skewed
things towards dakfu.  But "sword" has usage even in nonmedieval contexts.
Presumed approved.

dak *   *    25  6 dak          dakfu            25  19   1   1  12   6 knife
-   *   *     8 15              dakli dak        20  12   1   4   2   8 sack
----- net benefit 12

Lojban usage
dakfu
2    baldakfyxai
2    cladakfyxai
19   famdakfu
2    krudakfyxai
2    tordakfyxai
2    xandakfyxai
dakli
1    dakcka
3    gapydakcifnu
1    sipydakli

Eaton data
dakfu
dakfu +co +darxi

TLI dictionary
dakfu 1 initial
dakli 2 initial


9. NSN questioned the justification for pemci, though others liked it.
There are no real losers, so I support the change a bit more strongly.
Presumed Approved.

pem *   *    24 10 pem          pemci            24  14   0   3   4  10 poem
pen *   *     0  0 pen     pe'i penmi pempe'i    62  26  10   7   2  36 meet
pes *   *     0  0 pes     pei  pensi penpei    107  58  12  20  18  49 think
pex *   *     1  1 pex          pesxu pes         8   7   0   1   4   1 paste
pel *   *     0  0 pel          pelxu pex        13  13   2   4   1   0 yellow
pez *   *     1  1 pez          pezli pel        13  12   0   2   5   1 leaf
pef *   *     0  0 pef          pe'a  pez         1   1   1   0   0   0 start
----- net benefit 14

Lojban usage
pemci
9    ninpemci
1    pemcypra
8    rampemcyzba
4    salpemci

Eaton data
pemci +srana

TLI dictionary has 1 usage of pemci, in initial position (for poet)

11.  NSN questioned pilka's usage in Lojban.  The rafsi in question is
pi'a.  All Lojban usage of pilka has been in final position.  I think
pilka is used for coverings that are an integral part of the item, while
gacri is intended for alienable coverings.  Presumed approved.

*   *   pi'a 23  0 pil     pi'a pilka pil        36  13   0   3   6  23 crust
*   xra -     0  0 pir xra      pixra pirpi'a    69  32   5  14   7  37 pictur
*   -   xai   0  0         xai  xrani xra        49  26   0  11   8  23 injure
*   *   xa'i  0  0 xac     xa'i xarci xacxai     65  27   8   8   7  38 weapon
xar *   -     0  4 xar          xanri xa'i       16  12   0   6   2   4 imagin
-   *   *     0  0         xa'a xatra xarxa'a    40  14   3   1   3  26 letter
----- net benefit 19

Lojban usage
2    jmepilka
2    rasyselpilka
Eaton data
te +pilka +catke
na'e +pilka +zbasu
dertu +se +pilka
jdari +pilka
jadni +pilka
palku +pilka
bongu +pilka
kerfa +pilka
bisli +pilka
taxfu +pilka
pilka +punji
pilka +litki
jirna +pilka
jirna +pilka
djacu +pagre +pilka
katna +kerfa +jibni +pilka
TLI dictionary 2 initial, 1 final


13.  JC counterproposed leaving frinu with fin, NSN counterproposed
leaving frinu with fi'u.  MS preferred Nick's.  At this point given
relatively low usage for finti in non-final position and Cowan's initial
position proposals for frinu (thoughg I might use parbi or pagbu for
some of these), I am supporting Cowan's alternative, which does give
support for the cfipu usages.

*   *   fi'u 14  0         fi'u cfipu            14  12   1   4   0   2 confus
---
fin *   fi'i 23  0 fin     fi'i finti fit        34  11   0   4   0  23 invent
---
fit *   -     0 15 fit          friti fi'i       23   8   0   1   3  15 offer
--- cycle on fit
-   *   -     0  2 -            frinu finfi'u     3   2   0   0   1   1 fracti
----- net benefit 20

Lojban usage
cfipu
1    cfipyri'abau
1    selcfipu          (frequency understated - used much in DC conversation)
frinu - none
finti
1    ci'arfinti
1    fitci'a
4    fitfi'i
1    jbofinti
9    nunfinti
8    plafinti
5    selfinti
11   xamfinti
2    xamselfinti

Eaton data
cfipu
cfipu +cinmo
cfipu +pensi
frinu, finti - none

TLI dictionary does not support either frinu or cfipu as gismu or lujvo;
finti has 1 final position usage, with the concept of 'create' being
'cnino zbasu' and/or 'larcu zbasu', and author covered by "pemci zbasu",
"prosa zbasu", "tarti finti" = innovator???, or the well-known to TLIers
"start-giver" (this is the title JCB gives to himself re Loglan).  All
of these would normally be subsumed under finti for Lojban.

15.  JC, NSN, MS heavily debated this change vs.  Alternate proposal A.
David Twery, who created most of the cinse lujvo that caused the
original to be favored over the alternate strongly preferred giving
cinmo the good rafsi, indicating that he had gone a bit overboard on
cinse compounds, most of which were not likely to see common use.  I
agree, and Cowan also was willing to go along with a switch to alternate
proposal A. (They did favor giving cinse the better rafsi 'cin', and I
have done so.)  However, examination of the supporting data for first
position of citno now leads me to support giving BOTH cit and ci'o to
citno, at the expense of cinta.  The reduction in hyphenation more than
outweighs the minimal proposed usage for cinta.  The supporting data is
lengthy, so I give Lojban usages only and summarize the rest.

The main problem with all this fiddling is that the result offers almost
no improvement in statistical coverage, unlike the original proposal.
But it may be that local data must overrule statistics here.

15. revised per alternate A and more analysis
*   *   *    11 11              cinla            14  10   1   6   0   4 thin
this one unchanged
cis *   *     6  0 cis          crisa            11  11   0   3   3   0 summer
cin *   *    21 21 cin          cinse cis        75  54  12  17  19  21 sexual
-   *   *     1 14              cinta cin        17  16   4   6   3   1 paint
*   cni -     0  0     cni      cinmo ci'o      156  77  12  41  17  79 emotion
*   *   ci'o 15  0 cit     ci'o citno cit        56  41   1  28   4  15 young
*   -   ci'i  0  0         ci'i cinri cni        35  19   1   8   4  16 intere
*   cki -     0  0     cki      ciksi ci'i       43  12   0   4   2  31 explai
*   -   ci'u  0  0         ci'u ckilu cki        84  15   2  10   1  69 scale
tid *   -     0  6 tid          tcidu ci'u       17  11   0   1   3   6 read
sid *   *     0  0 sid     ti'i stidi tidti'i    43  14   0   8   0  29 sugges
=   *   *     0  0     dju      sidju siddju     62  43   9   9  18  19 help
----- net benefit 2

Lojban usage
crisa none; Eaton 1 medial that would still need a hyphen, TLI dict: 3 initial
cinse
1    cinsycroxra
1    cispe'o
1    cispendo
1    cisri'aboi
1    cistai
1    cistanru
1    cistau
1    cistrikei
1    cisyjvo
10   cisyselpli
1    cornuncinse
1    degnuncinse
1    gircinse
1    kaznalcisyselci
1    kelcistri
1    malcistrikei
7    nalcinse
1    nalcispe'o
1    nalnuncinse
1    nalsukcisfra
5    nuncinse
1    nunciskei
1    nuncisyjikca
1    nungaxycinse
1    nunmo'ucinse
1    nunselbaicinse
1    pursukcisfra
1    samcinse
1    selcinse
1    sevycinse
1    sukcaicisfra
   Eaton 3 initial 2 medial, 1 final; TLI dict: 2 initial
cinmo
3    camci'o
3    ci'ocro
5    ci'onri'a
8    ci'ordu'e
8    ci'orja'o
8    ci'orkansa
1    ci'orma'o
1    ci'orsinxa
1    depci'o
3    jivyci'o
5    kazyci'o
3    nuncinmo
6    selcinmo
4    sepci'o
8    smaci'o
2    to'erci'o
   Eaton data 5 initial, 5 medial, 18 final; TLI dict: 7 init. 3 medial 4 final
citno
5    citkanba
2    citnanla
10   citnau
28   citni'u
6    citpendo
10   citrai
3    fetcitno
20   nimcitno
2    slacitcu'e
4    to'ecitno
2    to'ercitnau
9    to'ercitno
   Eaton data 2 initial 1 final; TLI dict: 3 init.
cinta
3    cinfai
   Eaton data 1 medial; TLI dict: 4 init.
cinla
8    caircinla
1    cinlycai
   Eaton data 2 initial; TLI dict. 1 init.


19.  MS liked gar for gapru, though he doesn't say why.  'gap' is the
more natural rafsi, and all others preferred the change.  Presumed adopted.


20.  JC was undecided on these, and MS shrugged.  Supporting data for
danti and darsi provided.  They are minimal since TLI had neither word.
danti covers a wide range of English words (arrow, bullet, ballistic,
gun, cannon, missile); darsi has a much more limited range:  audacity,
dare, chutzpah, 'guts'. darno incidentally gets its more natural rafsi.
Cowan agreed.  Presumed adopted.

dan *   *    20 11 dan          danti            20   9   0   3   3  11 projec
dar *   *     0  0 dar     da'o darno danda'o    49  43   7  22   7   6 far
-   *   *     1  7 -            darsi dar        10   9   2   4   0   1 audaci
----- net benefit 3

Lojban usage
danti
2    seldanti
23   terdanti (which is meaningless given the new place structure)
darsi none


21.  JC, NSN, MS all commented.  There was no support for alternate B as
proposed; it was dropped.  JC suggested something else instead using
makcu, but it really takes a GOOD usage to justify displacing the
enormously used cmalu from a CCV.  NSN noted that the data for curnu is
erroneous, being based on his errors.  From the standpoint of this
proposed change, I can see no difference in taking away those numbers -
curnu was merely a side beneficiary of the main thread of changes.
However, if people feel it appropriate, cu'u can be given over to cuntu
per change 76. which would then allow macnu to regain cnu as some have
asked for.  I have changed the numbers for curnu to reflect Nick's
errors.

Nick was unhappy about bacru losing cru.  I agree that it was close to
sacred.  But the benefit to the much more useful cumki and barda and
curmi cannot be ignored (not to mention bringing cu'i the rafsi to track
with the cmavo).  I suspect that most usages of bacru should be cusku
anyway.

The tradeoff between mamta and cmana is not as simple as it might seem.
Yeah, I'd like to see mamta keep ma'a.  But it already has mam, while
cmana would otherwise have no cmavo.  Thus the scoring tradeoff is the
13 for final position of mamta vs. 39 for cmana, nearly twice as bad as
the worst coverage other than this one.

*   *   cu'i 17  0 cum     cu'i cumki cum        58  41   4  21   9  17 possib
---
-   bra *    19  0     bra      barda bad       131 112  28  61  14  19 big
---
-   cru -     0  0     cru      curmi curcu'i    86  25   1  14   4  61 let
----- with side benefit to
cur *   -     0 (5)cur          curnu cu'u        9   4   0   0   0   5 worm
---
*   -   ba'u  0  0         ba'u bacru cru        87  25   1   8   7  62 utter
bad *   -     0 11 bad          bandu ba'u       23  12   1   4   3  11 defend
-   bra *    19  0     bra      barda bad       131 112  28  61  14  19 big
*   -   ca'a  0  0         ca'a cabra bra        87  13   3   4   3  74 appara
*   *   ma'a  0  0         ma'a cmana ca'a       46  34   0  23   4  12 mounta
*   *   -     0 13 mam          mamta mamma'a    52  39  11  17   3  13 mother
----- net benefit 12 (the 5 from curnu losing cu'u is counted on 76 below)

erroneous curnu data
2    dapcu'u
2    jgicu'u
1    seircu'u
1    selcu'u
10   sevycu'u
(this amounts to a net score of 11 in final position, which is reflected above.)


76.  JC NSN MS all agreed that macnu had more need for a rafsi than
cutne.  But the real tradeoff for such a rafsi assignment is vs. cuntu,
which covers a wide and useful semantic range including 'affairs', and
'business'.  JC proposed 4 initial position lujvo for macnu.  I also
discovered I was missing some TLI dictionary data, giving macnu at least
9 more score in initial position, more than enough to justify a rafsi.
Given the realization of problems with curnu data in 21. above, I favor
moving cu'u to cuntu, allowing cnu to return to macnu.  The following
reflects that proposal.

cut *   *     6  0 cut          cutne             6   6   0   1   0   0 chest
cur *   -     0  5 cur          curnu cu'u        9   4   0   0   0   5 worm
-   -   cu'u  3  0         cu'u cuntu cut        19  16   4   6   3   3 affair
*   *   *    13  0 -   cnu      macnu cnu        13   1  13   0   0   0 manual
----- net benefit 17

Lojban usage
cuntu
17   cutyzu'e
cutne, macnu: none

Eaton data
cuntu
ka +no +se +cuntu
se +cuntu +cenba
se +cuntu +zbasu
se +cuntu +mukti

TLI dict:
cuntu 1 final
macnu 9 initial
cutne: none


22.  NSN indicates that sim is sacred for him for simxu; MS agrees,
labelling 'six' as "icky".  I am inclined to wonder if such a distaste
for 'x' is malglico, recognizing of course that Mark speaks languages
that use 'x'.  I will go along based on the rather less malglico
argument that 'six' is lousy for hyphenation as compared with 'sim', and
simxu has 30 1st position usages.  The decision then is which of stici,
since, or snime is denied a rafsi.  Note that directional usages of
stici are likely to be reversible in lujvo (e.g northwest = westnorth).
The data which seems to favor snime and stici, follows, as well as the
presumed modification to the change based on retracting sim for simxu.

sic *   *    10  9 sic          stici            18   9   0   1   0   9 west
-   *   *     4  8              since sic        12   8   2   2   0   4 snake
*   *   *     0  0         si'e snime si'e       13  13   0   7   0   0 snow
*   *   si'u 14  0 sim     si'u simxu sim        56  42   8  22   6  14 mutual
*   *   *     0  0 six          sirxo six        10  10   0   7   0   0 Syrian
*   *   &     0 18 nid          snidu nidsi'u    23   5   0   1   0  18 second
----- net penalty -7, ignoring snidu (metric/culture), net benefit 11

Lojban usage
snime
6    si'erbi'e
since - none
stici
6    berstici
15   snanystici
1    sticymla

Eaton data
since
banli +jubme +since

TLI dict.
stici - 2 final
snime - 2 initial
since - no lujvo


24.  NSN took an interesting position on this one.  He said NO, but
indicated that this was "not an irrevocable NO", being based on trying
for optimal hyphenation.  MS agreed with Nick.  The interesting thing is
that Nick was MUCH more vehement on the parallel case re fetsi, #150,
which follows, in spite of the fact that in that case fetsi was actually
going to retain a final position rafsi even with the change.  Now I ask
why fetsi MUST have a monosyllable CVV when nakni, even if na'i is
retained, gets only a disyllable CV'V.  Methinks a double standard is
being raised.  JC points out in response that probably most lujvo with
fetsi or nakni in final position can reverse the order and put the
gender in initial position.  It is not necessary that gender in Lojban
be a suffix, especially when the prefix is more likely to be
recognizeable in having 2 consonants.  On the other hand, we KNOW that
monetary units will exist for nearly every cultural gismu in final
position, and this is NOT a particularly reversible lujvo (cent-ish
American-thing???) nalci, of course, deserves a rafsi - there is
certainly going to be use for 'wing' in both initial and final
positions.  Nick's willingness to give on nakni which would give it no
final position rafsi at all, weakens his case on fetsi such that Cowan
and I vote equally strongly for the change as proposed.

*   *   na'i 18  0         na'i nalci            18   7   1   2   0  11 wing
*   *   -     0  9 nak          nakni nakna'i    25  16   1   9   1   9 male
----- net benefit 9

150. NSN vehemently opposes
*   *   fei   0  0 fep     fei  fepni fepfe'i    40  10   0   3   0  31 cent
*   *   fe'i  0  0 fet     fe'i fetsi fetfei     53  42  15  23   0  11 female


25.  MS asked for supporting data on tcaci and tcadu.  It was close
pre-Lojban usage, but not since. Presumed approved.

Lojban usage
tcaci
4    tcaju'o
tcadu
6    jbeta'uxa'u
1    lunryta'u
2    pijyta'u
4    ralta'u
3    ta'urkarni
8    ta'urlumpu'o
2    ta'urtrurkamni
3    ta'urxa'u
18   ta'utru
7    ta'uvro
3    ta'uvru

Eaton data
tcaci
tcaci +se +zukte
mutce +xlali +tcaci
tcadu
barda +tcadu
tcadu +ralju
bartu +tcadu +zdani

TLI dict.
tcadu 4 init 3 final
tcaci 5 init 1 final


27.  NSN wants justification for sakci, enough to be moving sal from
salci.  The main justification is that salci needs something it can use
in final position.  I am actually more disturbed by having to weaken
slabu and lasna.  The data shows that the usage of sakci comes mostly
from the compendium of DT's sexual terms, but that lasna appears to be
well-enough served by la'a.  I leave it to you people to choose between
slabu and salci, with the edge to salci based on statistics if there is
no consensus otherwise.

sak *   *    11  3 sak          sakci            11   8   3   2   1   3 suck
sal *   *     4  4 sal          sakli sak        13   9   3   2   0   4 slide
-   sla *    16  0     sla      salci sal        46  30   1  17   7  16 celebr
*   -   sau   0  0         sau  slabu sla        49  38  12  14   4  11 old
sas *   -     0  9 sas          srasu sau        38  29   2  12   9   9 grass
las *   *     1  1 las          slasi sas        13  12   2   6   0   1 plasti
-   *   *     0  0         la'a lasna lasla'a    36  17   1   7   4  19 fasten
----- net benefit 15

Lojban usage
sakci
2    gaxliksakci
2    sakcypinji
6    sakcyselylumci
salci
10   ctisalci
1    glesalcydei
3    jbosalci
1    nungumsaldansu
29   nunsalci
8    nunsaldansu
5    pixsalci
3    sa'irsalci
3    salcti
10   saldei
2    salgei
1    salja'o
4    salpemci
5    salsanga
slabu
2    mlecyslarai
4    ninsla
2    slacitcu'e
1    sladi'e
8    slagle
2    slamidju
1    slasutsabdja
2    zmaslarai
lasna
24   kikla'a
11   la'arja'i
1    nunla'a
1    nunlasna
3    sfalasna
1    skola'a
1    terla'a
1    terlasna

Eaton data
slabu
xlali +slabu
xamgu +slabu
slabu +stuzi
slabu +ckilu +merli
lasna, sakci, salci: none

TLI dict:  lasna is neither gismu or lujvo; sakci exists only as the
limited semantics lujvo mouth-pull, salci is the lujvo
grand-respect-give, though there are three unrelated lujvo for
celebration:  amuse-time, happy-time, giver-happy-time.  Excellent
examples for the weakness of the TLI language and their lujvo-making.
slabu: 6 initial position


32.  NSN and MS don't see why the following.  The major beneficiaries
are the first three words, and nothing much is hurt by the change:
sraji is the only word that loses coverage and almost all of its uses
are initial position, where a CVC will serve well (though I note in the
data it sraji seems to have an affinity for being followed by words that
start with unvoiced consonants, so there may be a lot of hyphenation.)
There is a hidden benefit in that sumne did not get any rafsi, so panci
may come to carry an extra load for sensory lujvo, but people haven't
done much in this area yet (maybe in the coffeeshop???).  Presumed
approved.

sat *   *     8  0 sat          sakta             8   8   1   4   0   0 sugar
-   *   sa'e  8  0         sa'e satre sat        19  11   0   4   1   8 stroke
---
pan *   *    12  1 pan          panci            12  11   1   4   1   1 odor
san *   *     0  0 san          spano pan        15  15   0  13   0   0 Spanis
---
    sna       0  0     sna      sance sansa'e   185 122  10  86  20  63 sound
*   -   sa'i  0  0         sa'i sanli sna        33  10   0   1   2  23 stand
*   *   sai   0  0         sai  sanmi sa'i       44  24   1   6  10  20 meal
*   sra -     0  0     sra      sarji sai        53  34   3  10  14  19 suppor
raj -   *     0  1 raj          sraji sra        53  52  15  26   8   1 vertic
-   *   *     0  0         ra'i ranji rajra'i    36  27   4   8   7   9 contin
----- net benefit 24

Lojban data
sakta - none
satre
1    cmilysatre
2    jisysatre
1    riksurlysatre
1    tacysatre
1    vibytacysatre
panci/sumne
sraji
3    drasratse
1    sraca'a
141  srake'a
1    srake'aloi
10   sramudri
2    srasirdra

Eaton data
sakta - none
satre
satre +ctebi
denci +kansa +satre
panci
panci +pluka
xamgu +panci
sumne - none
sraji
sraji +clani
sraji +fenra
sraji +pluta
sraji +danmo +pluta

TLI dict.
sakta - gismu but not used
sumne/panci - the gismu is for panci with sumne a derived lujvo; 3 initial
satre - translated as darxi, or the lujvo prami-pencu
sraji - 4 initial position


33.  NSN disputes my desire for tarmi to have a BAI that matches it
closely.  Both tadji and tarmi have been sumti tcita dating back to the
TLI days when there were only a dozen BAIs; they are among the most
important and likely to be used.  Nick himself missed the significance
of tarmi in badbarda/brabra kevna, if people will recall the rivers of
rock metaphor discussion last year.  The statistics also show that tarti
and tarmi are among the most used gismu in lujvo, and I am concerned
that people will bring the wrong one to mind when they see the
cmavo/rafsi based on the inconsistency.  Mark argued against this kind
of change elsewhere, but this was has been severe enough to affect me
and Nora, and I'm happy to find out that the data justifies the change
that I wanted to see.


35.  ( and 78.)  NSN and MS weren't convinced that moi would be used in
lujvo (and for that matter mei in #78, wherein JC also objected.
However, people have used pamoi and pamei in lujvo as well as
standalone; it is a frequent mistake by beginners - one with little
statistical supporting data, of course, since there is no way to make
lujvo with these words at all now. mei is easy to demonstrate the need
for:  all concepts wherein singularity is significant will be -pavmei;
likewise pairs will be -relmei, and trios, quartets, quintets, dozens,
scores, grosses, abound in English and I presume other languages.  They
all need a mei rafsi, or instead must be tanru with the number selbri
probably in final position, which usually gives a useless place
structure.  "moi" is a little harder to come up with examples for, but
"first" and "last" -pavmoi and -rolmoi seem especially productive.
Orchestras have first and second violins, and Loglan/Lojban now has a
first-digger (pavmoikakpa).  Ordinals are less used in natlangs than
cardinals, but when they are used, their semantics seems to stress the
order more than the number itself.  I think the numerical selbri
standing apart, on the other hand, tend to stress the number, and not
its semantic usage.


37.  NSN wanted the assignments changed, even at the expense of toldi,
to suit some apparently aesthetic consideration.  I distrust aesthetics
as a motivation for anything in Lojban as being inherently culturally
biased, but this one needed changing anyway since people decided number
rafsi were sacred, and toldi can pick up the 'overkill' 'tod' from
stodi.  So in this case I can oblige Nick's aesthetics.

*   *   *     0  0 non          no    non         7   7   2   3   2   0 0
unchanged
tol *   *     0  0 tol     to'e to'e  to'e       45  45   4  35   6   0 polar
nor *   *     0  0 nor     no'e no'e  no'e       11  11   4   7   0   0 neutra
tod *   -     0  2 tod          toldi to'i        6   4   0   0   0   2 butter
=   *   *     0  0     sto      stodi todsto     12   9   1   3   1   3 consta


38. JC NSN MS all agreed that 2 rafsi for romge was excessive.  I won't argue
in the face of massive rafsi minimalist semtiment, and have dropped ro'e.

rom *   roi   0  0 rom     roi  roi   rov         0   0   0   0   0   0 quanti
rog *   -     0  0 rog          romge romro'e     3   3   0   1   0   0 chrome
*   *   ro'i  0  0 rok     ro'i rokci rokroi     59  33   1  13   8  29 rock
toc *   toi   0  0 toc     toi  troci rocro'i    41  22   2   7   4  19 try
*   *   to'i  0  0 ton     to'i torni tontoi     17  14   3   5   0   3 twist
tod *   -     0  2 tod          toldi to'i        6   4   0   0   0   2 butter
----- roc is freed and is not useful to any other word
----- rog is currently unassigned
----- net penalty 2
----- roi proposed by me as needing a non-hyphenating rafsi


39.  JC NSN MS all agree that gocti and gotro do not need the CVV rafsi
(goi and go'o), and they are deleted from the proposal.  The new gismu
were approved at LogFest.  Giving bep to zbepi is overkill; it AND ze'i
are dropped in the revised proposal.

goc *   *        0 goc          gocti                                   1e-24
zep *   *        0 zep          zepti                                   1e-21
got *   *        0 got          gotro                                   1e24
zet *   *        0 zet          zetro                                   1e21
=   *   =     0  0     zbe      zbepi zepzbeze'i  9   3   0   1   1   6 pedest


40.  Correcting a historical note:  Cowan thought that pante was a new
gismu, which it is not.  It dates to the TLI era, though it has only 1
final position lujvo usage in the TLI dictionary.  In any case, the
proposal is approved since not opposed.


42. JC was undecided on this one.  Here is the data.

mat *   *    16  3 mat          mapti            16  13   0   6   1   3 fit
-   *   *     1  6 -            matra mat         6   5   2   3   0   1 motor
----- net benefit 8

Lojban data
mapti
1    maptybi'o
3    maptypai
2    maptype'i
matra-none

Eaton data
mapti-none
matra
matra +trene

TLI dict.
matra - 1 init. 1 final
matci is reflected in the brain-dead lujvo "tarmi-mintu", as if things
must always be of identical shape to match each other.


43.  NSN shrugged; the others said yes.  Here's the data showing a
preponderence of people talking about the weather in final position,
especially in recent times; I'll assume it approved.

*   *   ti'a 15  0 tim     ti'a tcima tim        35  20   4   5   6  15 weathe
*   *   -     0 14 tic          tcica ticti'a    48  34   9  11   8  14 deceiv
----- net benefit 1

Lojban data
tcima
22   cictcima
15   viltcima
1    xlatcima
tcica
3    maltcica

Eaton data
tcima
nu +tcima +dikca
vlile +tcima
tcima +pruce
vlile +tcima +simsa
klina +tcima +cenba
tcica
tcica +se +simlu
tavla +tcica
jitfa +tavla +tcica

TLI dict.
tcima - 2 init. 1 medial, 1 final
tcica - 1 init.


47.  MS wanted justification on this, but the others approved.  There is
inherently not going to be a lot - given that rijno is rather limited
semantically, and hence is not likely to be too productive.  I won't
argue too hard if people want to reverse this one.

rij *   *    12  0 rij          rijno            12  12   1   5   0   0 silver
-   *   *     0  0         ri'u rinju rijri'u    43  23   4   4  10  20 restra
----- net benefit 12

Lojban data
rijno
1    rijnysi'a
2    rijnyska
rinju
8    ri'usrutu'o
1    selri'u

Eaton data
rijno
rijno +simsa
rinju
rinju +cabra
cabra +rinju
zukte +rinju
tarti +rinju
stuzi +rinju
muvdu +rinju
rinju +javni
botsu +rinju +genxu
muvdu +sevzi +rinju
genxu +rinju +bloti +stuzi
cabra +rinju +botsu +stuzi

TLI dict.
rijno 2 initial (including rijno simsa again)
rinju 5 final


48.  NSN saw no particular justification for gerku, except for
figurative ones.  It does seem to have been bolstered by the iterations
of Open Window in the files, but there seems to be plenty of other
usages even if some are quite figurative.  I consider the gleki article
in JL to be exactly what I DON'T consider to be sacred.  All of these
statistics are based on such spewing out of numerous proposals, and many
of them have seen print in either JL or TLI publications.  To me, a
rafsi is sacred if people know a particular lujvo based on it well
enough to use without analysis (in which case they won't notice until
they confuse someone that the meaning has been changed out from under
them), or something that we have used in teaching documents extensively
enough that people would notice the change (especially if we missed
revising it, a la kunbri - by my standards, kun would have been sacred
is this review based on that usage, even though the metaphor was lousy.
I have never seen any indication that people even READ that gleki game,
much less used it as a source of lujvo, or as intended, tried to do the
same for some other semantic field.)

ger *   ge'u 12  0 ger     ge'u gerku gek        32  20   0  12   0  12 dog
---
gen *   *     0  0 gen     ge'a gerna gerge'a    43  11   3   0   1  32 gramma
=   *   *     0  0     jge      jgena genjge     18  12   0   9   3   6 knot
---
*   gle -     0  0 let gle      gletu letge'u    18  13   1   2   4   5 copula
gek -   *     0  0 gek     gei  gleki glegei     83  51   5  25  13  32 happy
----- cycle back on gek
----- net benefit 12

Lojban data
3    gekcru
6    gekpre
8    gekyki'a
1    jvigerku
9    malgerku
1    nungerku
24   pangerku
1    simygerku

Eaton data - none
TLI dict:  2 initial (nakni/fetsi)


51.  NSN thought taj sacred due to tajnau.  JC answers that he would not
want the language bound by people's earliest Lojban writings - he would
not have used that lujvo if he were writing the same text today.  MS
asked re justification for tamji, and I'll give non-final uses of traji
as well.

taj *   *    10  2 taj          tamji            10   8   0   2   2   2 thumb
-   *   *     0  0         rai  traji tajrai    210  63   4   8  43 147 superl
----- net benefit 8

Lojban data
traji
10   tajnau
tamji
2    jaftamji
2    jaftamjycalku

Eaton data
tamji
tamji +clani
tamji +ganra
traji in TLI Loglan is only expressed as "zmadu roda"

52.  NSN gives a lujvo for maze "lujypludi'u" that would lose a hyphen
by this proposal "lujlu'adi'u", and claims that pluta will be more used
in final position.  The latter is true as the data below shows, but
affects decision-making only if you buy the argument that CVVs are poor
in final position, which we've already argued to no consensus - the data
goes both ways, and it probably depends on aesthetics as well as whether
you are writing or speaking the word.  Since even those opponents of CVV
final use them in final position when they can use the expanded form as
well, I think this is intellectual rather than pragmatic aesthetics that
we are dealing with.  In any event there is clearly significant use of
daplu, most of it final, and pluta does remain covered.

daplu
4    badydaplu
1    cmadaplu
2    xabdaplu
pluta
3    di'uplu
18   dijyplu
2    dzuplu
1    jacplu
8    kacplu
39   lujypludi'u
3    plufa'o
1    pluke'a
1    pluku'a
8    velplu

Eaton data
daplu
pagbu +daplu
xadba +daplu
pluta
pluta +cirko
sraji +pluta
flecu +pluta
tricu +pluta
danmo +pluta
cliva +bartu +drani +pluta
bartu +pluta +cliva
pluta +sisti +rinka
sraji +danmo +pluta
ciblu +fatne +pluta
jdari +rinka +jamfu +pluta

TLI dict.
daplu is a gismu, but not in any lujvo
pluta 2 initial, 6 final


55.  MS asks whether creka needs a final form - the data shows only one
usage in that position.  He and JC agree that cet should be freed, so
the proposal is so amended.

*   *   cei   4  0 cev     cei  cevni cev        58  54  16  28   2   4 god
---
*   *   ce'u  9  0 cem     ce'u cecmu cem        22  13   2   5   0   9 commun
=   cre -     0  0     cre      certu cetce'u    61  19   2   9   2  42 expert
cek -   *     0  1 cek          creka cre         8   7   0   2   1   1 shirt
---
tek *   ce'i  0  0 tek     ce'i cteki cekcei     14   9   1   4   2   5 tax
*   *   -     0  2 teb          ctebi tebce'i     9   7   1   1   1   2 lip
----- tek was previously unused
----- net benefit 10


59.  NSN questioned final position usage of renvi in this change.  All
lujvo usage thus far has been in final position, and JC notes that
"survivor" is a form that may suggest more usage (also victim, and outlast,
in certain contexts are renvi).

*   *   re'i 10  0 rev     re'i renvi rev        23  13   2   5   1  10 surviv
*   bre -     0  0 red bre      bredi redre'i    35  29   3   9  11   6 ready
bes -   be'a  0  0 bes     be'a bersa bre        42  12   1   6   0  30 son
jer *   -     0  3 jer          jbera be'a        8   5   0   2   0   3 borrow
-   *   *     2  2 -            jerna jer         6   4   0   0   0   2 earn
ben *   *     0  0 ben          besna bes        25  25  10  10   0   0 brain
=   *   *     0  0     jbe      jbena benjbe     84  59  19  24   9  25 born
----- net benefit 7

Lojban data
2    ninrenvi
9    nunrenvi
4    selrenvi

renvi was not a TLI gismu or lujvo


64.  JC was strongly for this one; NSN MS less so.  Rice has some
obvious compounds in food discussions; probably more in cultures where
rice is more important (The Chinese may have 57 words for rice, instead
of snow ^) risna is useful in medicine, where the adjectivial form
cardiac may suggest some lujvo, but the 2 lujvo usages actually recorded
are from Eaton data, and are tied to the malglico "broken-hearted".
.ionaicai  Presumend approved.

ris *   *     9  0 ris          rismi             9   9   1   0   4   0 rice
-   *   *     2  2 -            risna ris         8   6   0   0   0   2 heart
----- net benefit 9


65.  NSN MS agreed that this was an unjustified clunker.  The supposed
beneficiaries were murse and sorgo which casual inspection showed had no
real usage.  This change was an artifact of my early statistical methods
that should have been caught when I changed them in midstream, and had
no justification.  Cowan argued for deletion of at least one of the
rafsi for sorcu on general principles, though, and I consented.

Revised:
*   *   =     0  0 soc sro      sorcu socsroso'u 69  42   7  14  16  27 store
murse, muslo, solji, so'a, sorgu unchanged
----- net change 0


68.  NSN is opposed, being suspicious of lujvo involving ladru.  Why?
Have you never seen a dairy farm, eaten or drunk dairy products, etc.
Then there are specific products:  goat-milk-cheese, etc.  And this even
presumes that the concept "milky" is impermissible as a description of
appearance for a liquid.  (I won't argue for "Milky Way").  Whether
these outweigh cladu's needs may be arguable, but the usages are there.
JC and I are strong yes, MS a weak no.  I can't argue if Nick says ladmau is
sacred to him

lad *   *     8  0 lad          ladru             8   8   0   3   0   0 milk
-   *   *     0  0         lau  cladu ladlau     41  31   2  18   4  10 loud
----- net benefit 8

Lojban data
cladu
2    cladakfyxai
2    ladbacru
2    ladbi'o
4    ladmau
1    ladyckasu
8    laucru
1    nalcladu
2    to'erlau
ladru - none

Eaton data
cladu
zanru +cladu
cladu +darlu
so'i +cladu +sance
farlu +ja +porpi +cladu
ladru
ladru +dinju

TLI dict.
cladu 3 init. 1 medial 2 fianl
ladru 1 init (milky)


72.  MS thought this one made no sense, and on second look at the
statistics, I agreed.  My standard was to divide the number of 1st
position usages by 4 or 5, and even this gives jundi more need than the
single usage of judri non-final.

                                judri             7   6   0   0   1   1 addres
                                jundi judju'i    42  20   1  12   0  22 attent
no change



81.  JC, NSN, MS agreed that staku was more useful than taske in lujvo.
Neither had seen any use in lujvo, but taske had been used as a gismu
several times, unlike staku.  Given Cowan's actual proposals for lujvo,
the proposal was withdrawn.

                                taske             4   4   0   0   0   0 thirst
                                staku tak         1   1   0   0   0   0 cerami
no change


84.  NSN MS both oppose this change, MS less strongly.  JC and I both
support it.  (JC also wants 'ded' dropped as overkill, and I agree, and
have modified the proposal.)  The tradeoff is between supporting denci,
or labelling a rafsi that probably should not have been used EXCEPT for
malglico reasons as sacred.  Both opponents consider it vital to have a
rafsi for dinri, yet MS says that all the lujvo made with 'dei' are
djedi based and not dinri.  If so, then there is no usage to justify a
rafsi for dinri; one cannot have it both ways.  Both denci and degji
will be used in final position, both literally and in the same kinds of
shape-related figurative metaphors.

This one really seems to be the big vote on sacredness vs. efficiency.
So lets see if the data changes any votes or leads to a consensus.

The dinri problem has been solved - based on the ckamu, mleca precedent,
JC and I proposed changing the gismu to give it good rafsi, arguing that
as a newly created word it is not yet well-known, and the extreme demand
for good rafsi for it (though almost unjustified by actual examples, I
must admit) warrants overriding the sanctity of the gismu list.  This
was approved at LogFest, and CF also agreed that changing the gismu was
acceptable in this case.  (Sorry if this sounds like a railroad, but we
didn't have much time to check with people between proposal and vote -
Colin happened to send a message at the right time.)

I want people to go through the lujvo data below and determine which if
any should be switched to donri.

*   *   de'i  4  0 den     de'i denci den        28  24   2   8   9   4 tooth
*   *   dei   0  0 deg     dei  degji degde'i    31  23   3  11   5   8 finger
=   *   =     0  0     dje      djedi deddjedei 102  39   9  17   5  63 full d
dor *   do'i       dor     do'i donri [dinri]     4   4   0   0   0   0 daytime
----- net benefit 4

Lojban data
denci
26   xantydenmai
1    dentro
degji
2    badjafyde'i
2    badjafyde'icalku
2    badyde'i
4    degjai
1    degnuncinse
2    pavde'i
djedi
11   bavlamdei
2    bendei
30   cabdei
13   dedmidju
1    djesni
1    djetei
1    glejbedei
1    glesalcydei
3    midydei
2    midydeisa'i
8    nacykefydei
1    pavnondei
9    prulamdei
11   purlamdei
4    roldei
10   saldei
1    zandei

Eaton data
denci
denci +bongu
denci +danlu
degji +denci
barda +denci +marji
denci +kansa +satre
jamfu +degji +denci
degji
degji +denci
degji +sinxa
jamfu +degji +denci
djedi
pa +djedi
pa +djedi
re +purci +cabna +djedi
djedi +krasi
gleki +djedi
jbena +djedi
morji +djedi
djedi +liste
djedi +xriso +re +moi +jbena
djedi +morji +cukta
ciste +djedi +nanca

TLI dict.
denci 3 init.
degji 5 init; also toe was separate gismu with no lujvo
djedi (their djedi is defined as our dinri/donri, though you'd never guess it
       from the lujvo)
 xriso djedi
 midju djedi
 lamji djedi
 ro    djedi
 purci djedi
 jbena djedi
 djedi nicte (our djedi)
 djedi sinxa
 balvi lamji djedi
 purci lamji djedi
 djedi sinxa ciska

85.  JC labelled this one "no winners, only losers".  I asked him what
this meant.  His philosophy is to try to give every gismu at least one
rafsi to some extent regardless of usage.  Thus zbani loses. zarci
doesn't win by this philosophy since it already has a rafsi, even though
any number of types of stores and marketplaces will use zarci in final
position.  JC now supports the proposal, but I list supporting data for
zasti and zarci, and people can decide whether they want to reverse zai
and za'i between these two.

*   *   za'i  3  0 zac     za'i zarci zac        15  12   0   2   1   3 market
*   *   zai   0  0 zat     zai  zasti zatza'i    30  27   3  10   7   3 exist
*   *   -     0  2 -            zbani zai         5   4   0   1   0   1 bay
----- net benefit 1
----- The statistics don't support this very strongly, but I can imagine
countless numbers of lujvo for types of markets/stores/malls, and few
for bay.

Lojban data
zarci - none
zasti
1    cabnalzasti
1    za'irbi'o
3    zatsta
zbani - none

Eaton data
zarci
ponse +pagbu +zarci
pagbu +ponse +zarci
pagbu +ponse +zarci +stuzi
zasti
zasti +sisti
so'i +zasti

TLI dict.
zarci - 2 init.
zasti, zbani: no lujvo for these gismu


86.  MS thought virnu was more likely to be initial, apparently ignoring
the large numbers in final position (which are real - many of them date
from my own TLI dictionary work before Lojban), and that the vri could
be assigned somewhere more valuable.  There aren't all that many words
that have a better excuse for CCV than one with 11 initial, 9 medial,
and 22 final in score, and besides - no other gismu COULD be assigned
vri (including vidru, even if we had wanted to - it would qualiy only
for vid, vir, vi'u, and dru).  Presumed approved.

vir *   *     3  0 vir          vidru             3   3   0   1   0   0 virus
=   *   *     0  0     vri      virnu virvri     48  26   2   9   9  22 brave
----- net benefit 3


87.  NSN and MS both considered these poorly justified, with Mark
labelling it a lot of clunking in pursuit of a promise of cmavo
correspondence that is unachievable.  I'm convinced.  Proposal
withdrawn.

                                junta             2   2   0   2   0   0 weight
                                jutsi jut         7   6   0   0   1   1 specie
                                jursa jus         7   4   0   0   0   3 severe
                                junri jur        11   8   0   2   2   3 seriou
                                djuno junju'o    99  60  12  20  19  39 know
                                dunli dundu'i    50  34   0  10  17  16 equal
                                jduli duljdu     11   8   0   7   0   3 jelly
                                du    dubdu'o     2   2   2   0   0   0 same i
                                du'u  dum         0   0   0   0   0   0 bridi
                                jungo jug        16  16   1  11   0   0 Chines
unchanged



90.-92.  NSN MS questioned whether these changes were justified.  In
90., it is a toss-up, since neither is used in final position.  But
snura clearly has much more overall use and can thus is more likely to
make use of it.

*   *   nu'a  2  0 nur     nu'a snura nur        52  50   9  29   6   2 secure
*   *   -     0  1 nuz          nuzba nuznu'a    24  23   0  13   3   1 news
----- net benefit 1

Lojban data
nuzba
4    nu'arki'a
8    nuzyxagji
snura
8    nurcau
12   nurgau
2    nurprecau
4    nurpu'i
2    nurstapa

Eaton data
nuzba
turni +nuzba +vasru
snura
nu +snura +zbasu
na'e +snura +krici
snura +krici
snura +rinka
snura +kurji
snura +taxfu
snura +zbasu +damba
cabra +zbasu +snura +ja +mipri

TLI dict.
nuzba 3 init.
snura 4 init.


91.  In this case, not a lot of justification is needed, since the
changes are primarily due to freeing up 'overkill rafsi' JC argued and
convinced me that korcu does not need koc and it is dropped in this
iteration.

kob *   ko'i  2  0 kob     ko'i kobli             2   2   0   0   0   0 cabbag
*   *   =     0  0     sko      skori skoko'i    31  16   3  10   1  15 cord
kor *   *     0  0 kor     koi  korbi kobkoi     41  20   1   9   4  21 edge
=   *   *     0  0     kro      korcu korkro     26  20   2  10   4   6 bent
----- net benefit 2

92.  In this one, it seems clear that derxi, though thus far little
used, will be used primarily in final position - there are many kinds of
heaps, but few heapish things.  For the aesthetes, a rafsi with no 'x'
may increase its use.  Overkill justifies some change in dertu; the
change proposed supports derxi at the cost of a CVV on little used
desku.  I argue on the basis of projected use, and not proven
productivity which is indecisive.  JC argues based on overkill, that dex,
should be deleted, and the proposal is modified.

=   dre *     2  0     dre      derxi dex         6   4   0   0   0   2 heap
*   -   de'u  0  0 der     de'u dertu derdre     64  41  11  16   7  23 dirt
*   *   -     0  1 des          desku desde'u     9   8   1   0   1   1 shake
----- net benefit 1

Lojban usage
derxi
1    maurderxi
desku
1    cmade'u

TLI had lujvo for both of these, desku being a family of compounds of
slilu, and derxi being either marji-cmana or marji-galtu-cmana depending
on how big the pile was.

94. MS and NSN don't see much point.  There is indeed a minimal justification, but
on later thought, given only a 1 point gain vs. the unnatural and more hyphenating rafsi for cunso, I'm
withdrawing this one.

                                kusru kus        10   9   2   2   0   1 cruel
                                ckunu ku'u        0   0   0   0   0   0 conife
                                cunso cuncu'o    70  65  20  28  12   5 random
                                cusku cussku    326  70   7  22  31 256 expres
unchanged


95.  JC MS NSN - 2 shrugs and a NO, but JC became undecided at LogFest
review.  Here is the data.  Please revote.  I won't fight either decision.

*   *   di'o  7  0 dig     di'o dirgo dig        15   8   4   0   0   7 drop
*   *   -     0  7 -            dinko di'o        7   3   0   1   2   4 nail
----- net benefit 0

Lojban data
dirgo
8    digligykamju
1    jacdirgo
5    klakydirgo
dinko
2    cardi'o

Eaton data
dirgo
cerni +dirgo
dinko-none

TLI dict.
dirgo - gismu but no lujvo
dinko 1 init, 1 mid, 2 final (a good lesson in TLI semantics:
  cmalu dinko = tack/brad
  cmalu dinko gasnu = to tack
  dinko gasnu = to nail
  barda dinko = spike)

Nora mentioned that 'staple' as a noun and verb would also use dinko.


97.  JC noted an error; moklu was omitted from this list as picking up
'mok'.  But I am having second thoughts - the choice is between the more
natural 'mok' and the rarely hyphenated and status quo 'mol'. 97 is
presumed approved.  Please vote on 97a, with my leaning towards the
status quo.

mov *   *        0 mov          mo'i
mor *   *     1  1 mor          morko mok        10   9   0   7   1   1 Morocc
mon *   *     0  0 mon     mo'a morna mormo'a   129 110   4 101   1  19 patter

97a.
*   *   *     0  0 mok     mo'u moklu molmo'u    23  24   1  12   2   4 mouth


98.  JC NSN MS all supported jve for je, and opposed all others.  Easy
to agree with.

Modified:
                                jo    jov         0   0   0   0   0   0 tanru
                                ju    juv         0   0   0   0   0   0 tanru
                                javni jva        73  45   5  18  18  28 rule
                                jei   jez         0   0   0   0   0   0 truth
                                ja    jav        11  11   0   0  11   0 tanru
*   jve *     0  0 jev jve      je    jev         5   5   0   0   5   0 tanru


99.  JC NSN MS all agree that extra rafsi for ce are overkill, but the
goal was to give a better rafsi to cecla, which has actually been used,
as opposed to the only theoretical usage of ce.  Per JC suggestion then,
all but the CVC for ce has been removed from the proposal.

cel *   *     0  0 cel     ce'a cecla cecce'a    28  12   0   7   2  16 launch
cec *   *     0  0 cec          ce    cel         0   0   0   0   0   0 set co
----- net benefit 0


100.-104.  JC NSN MS all agreed that number rafsi were sacred, also
affecting 4, 7, and 0 elsewhere in the set of changes.  All are
withdrawn.  The other less-hyphenating changes ofr 101-103 were
acceptable to all, and hence adopted.

jom *   *     0  0 jom          jo'e  joz         4   4   0   0   4   0 union
pus *   *     0  0 pus          pu'i  puz         0   0   0   0   0   0 can an
tuf *   *     0  0 tuf          tu    tuv         0   0   0   0   0   0 that y
                                bi    biv         0   0   0   0   0   0 8
                                xa    xav         1   1   0   1   0   0 6


105.  NSN argued that cokcu doesn't need cok either, and this is more or
less agreeable.  Proposal modified.

*   *   co'u     0         co'u co'u
=   *   =     0  0     cko      cokcu cokckoco'u  4   3   0   1   0   1 soak

106.-134.  JC NSN MS were generally opposed 'modulo exceptions'.  We
went thru these individually, and generally deleted 1 or 2 rafsi unless
we really though there would be a lot of possible use.

106.
bus *   bu'i     0 bus     bu'i bu
 approved

107.
col *   *        0 col          co
 modified

108.
*   *   co'a     0         co'a co'a
 modified

109.
                                go'i
 rejected/withdrawn

110.
*   *   jo'u     0         jo'u jo'u
 modified

111.
mob *   *        0 mob          mo'a
 approved

112.
*   *   *     0  0         ce'o ce'o  ce'o        0   0   0   0   0   0 sequen
 status quo

113.
com *   *     0  0 com     co'e co'e  co'e        2   2   0   2   0   0 unspec
 approved

114.
*   *   doi   0  0 don     doi  do    don         5   5   1   4   0   0 you
 approved

115.
*   *   *     0  0         fo'i fo'i  fo'i        0   0   0   0   0   0 it-8
 status quo

116.
lem *   *     0  0 lem          le'e  le'e        0   0   0   0   0   0 the st
 modified - these will probably allways occur in 1st position

117.
lom *   *     0  0 lom          lo'e  lo'e        0   0   0   0   0   0 the ty
 modified - these will probably allways occur in 1st position

118.
*   *   *     0  0 nun          nu    nun       355 355  97 215  43   0 event
 status quo

119.
*   *   *     0  0         nu'o nu'o  nu'o        0   0   0   0   0   0 can bu
 status quo

120.
*   *   *     0  0 sel          se    sel       880 880 136 460 284   0 2nd co
 status quo

121.
sup *   *     0  0 sup     su'e su'e  su'e        0   0   0   0   0   0 at mos
 approved

122.
suz *   *     0  0 suz     su'o su'o  su'o        2   2   2   0   0   0 at lea
 approved

123.
*   *   *     0  0 ter          te    ter       176 176  26 120  30   0 3rd co
 status quo

124.
*   *   *     0  0 vel          ve    vel        47  47   8  38   1   0 4th co
 status quo

125.
*   *   *     0  0         ve'e ve'e  ve'e        0   0   0   0   0   0 unspec
 status quo

126.
*   *   *     0  0 von          vo    von         4   4   3   1   0   0 4
 withdrawn - sacred

127.
*   *   *     0  0 vuz          vu    vuz        13  13   8   0   5   0 yonder
 status quo

128.
*   *   *     0  0 xem          xe    xem         0   0   0   0   0   0 5th co
 status quo

129.
*   *   *     0  0         za'o za'o  za'o        0   0   0   0   0   0 superf
 status quo

130.
*   *   *     0  0 zel          ze    zel         0   0   0   0   0   0 7
 withdrawn - sacred

131.
*   *   *     0  0         ze'e ze'e  ze'e        0   0   0   0   0   0 unspec
 status quo

132.
zev *   *     0  0 zev     ze'o ze'o  ze'o        0   0   0   0   0   0 recedi
 approved

133.
zon *   *     0  0 zon     zo'a zo'a  zo'a        0   0   0   0   0   0 tangen
 approved

134.
zor *   *     0  0 zor     zo'i zo'i  zo'i        0   0   0   0   0   0 approa
 approved


135. JC NSN MS all said this was overkill.  Agreed at Logfest to drop
both CVCs, retain foi because of some past usage.

*   *   *     0  0         fo'a fo'a  fo'a        1   1   0   0   1   0 it-6
*   flo *     0  0     flo foi  foldi folfoi     41  12   4   3   0  29 field


136. JC NSN MS agreed that this was overkill, CVCs deleted.
*   *   *     0  0         fo'e fo'e  fo'e        1   1   1   0   0   0 it-7
=   *   *     0  0     fro      forca forfro     10   6   0   1   0   4 fork

137. JC NSN MS agreed that this was overkill - deletions made
*   *   *     0  0 vaz          va    vaz         4   4   0   4   0   0 there
=   *   =     0  0     vra      vraga vagvrava'a 16   7   1   2   1   9 lever

138. JC felt diz was overkill, but accepted it as natural CVC from status quo.
*   dzi *     0  0 diz dzi      dizlo diz        14  14   0   7   0   0 low
*   =   *     0  0 zip     zi'o dzipo zipdzizi'o  3   3   0   0   0   0 Antarc

145.  NSN labelled this whole region for overkill, though John hadn't
labelled this one, which was moving an 'overkill' rafsi from a less-used
to a more-used gismu.  But on mixre, the extra rafsi is the natural one,
and actually might improve hyphenation.  No such benefit applies to
milxe, so it probably wouldn't be used.  Thus, by NSN's general comment,
this one is withdrawn.

                                milxe mli        41  28   1  20   1  13 mild
                                mixre mixxre     25  19   2   4   8   6 mixtur
no changes

146.  MS shrugged, but there really is a lot more non-final trixe than
rirxe, and the one place rix might have been used apparently fell victim
to aesthetics.  Presumed approved.

rix *   *     0  0 rix     ti'e trixe ti'e       57  45  11  23   6  12 behind
-   *   *     0  0         ri'e rirxe rixri'e    40  28   8  13   0  12 river

Lojban data
trixe
7    cucti'e
10   jafti'e
1    nuntrixyga'abikla
2    ti'erbakfu
2    ti'erbongu
2    ti'erkansa
1    ti'ertu'e
8    trixystu
rirxe
22   cmari'e
8    ri'emla
2    ri'erkrajinto

Eaton data
trixe
nu +trixe +muvdu
jamfu +trixe
trixe +farna
trixe +catlu
trixe +claxu +klama
trixe +cliva +prenu
trixe +farna +klama
rirxe
rirxe +moklu
rirxe +ganxo
suksa +rirxe
rirxe +sisti

TLI dict.
trixe 9 init 3 mid, 3 final
rirxe 3 init 2 final


147.  JC NSN thought sem was overkill, but ignored that te'o on terto is
even more likley overkill.  Modified to only drop extras on semto, others
are status quo.

                                terto tet        77  77  68   9   0   0 1E12
                                stero te'o       19   3   0   1   0  16 sterad
=   *   =     0  0     sme      semto semsmese'o  6   6   0   5   0   0 Semiti

148.  MS was worried about robbing kelvo of 'kev'.  The only use
non-final that I can think of is for temperature/thermometer.  Given how
little known/understood kelvo is in the first place, I doubt that many
will even consider kelvo for lujvo-making.


149.  JC NSN argued overkill on bolci.  I agree to drop bo'i, but
believe that 'bol' deserves retention as natural status quo CVC, and is
at least as justified as 'dei' for djedi, especially since there is no
competition.  JC agreed to the proposal as modified.

*   -   loi   0  0 lot     loi  bloti lotblolo'i 81  54  18  21   9  27 boat
*   blo *     0  0 bol blo      bolci bolboi     87  51  15  23   7  36 ball
*   *   boi   0  0 bot     boi  botpi botbo'i    13  12   2   3   0   1 bottle
*   *   lo'i  0  0 lol     lo'i loldi lolloi     44  34   7  14   6  10 floor

154.  MS argued for sacredness of gir for girzu.  JC and NSN let this
slip by their overkill sensors, for some reason.  Based on both
arguments, the entire proposal is withdrawn.

                                jgira jgi        26  16   2   9   2  10 pride
                                girzu girgri    197  76  27  17  25 121 group


158. JC NSN argue overkill. Proposal modified.
*   *   *     0  0     kre      kerfa kre        67  36   4  20   6  31 hair
ref *   *     0  0 ref     ke'u krefu kefke'u    75  65  11  33  15  10 recur
----- ref is currently unassigned

159.-161.  JC NSN MS all consider these a waste and overkill.  I'm not
sure since each offers possible alternatives in hyphenation situations
in words that have much lujvo usage (well, except maybe petso, which hs
only theoretical usage).  I won't argue too hard.  Withdrawn
reluctantly.

lip *   *     0  0 lip vli      vlipa vli        54  37   4  21   6  17 powerf
*   tso *     0  0 pet tso      petso pet        77  77  68   9   0   0 1E15
tuc *   *     0  0 tuc ctu      ctuca ctu        53  26   3  10   2  31 teach




201.
=   *   *     0  0     bru      burcu bucbru     17  11   1   9   0   6 brush

202.
=   *   *     0  0     cne      cenba cebcne    256 125   4  20  95 131 vary

203.
*   *   *     0  0 con cno coi  condi concnocoi  30  30   9  12   3   0 deep

not changed; semi-sacred - used in synopsis lujvo-making examples which
depend on having a word with all 3 possibilities - I think it worthwhile
to keep this one case of a word with all 3 rafsi, even if just for the
purpose of making examples.

204.
=   *   *     0  0     fle      flecu fecfle     77  32   5  11   9  45 flow

205.
=   *   *     0  0     fli      fliba fibfli     42  21   3  13   2  21 fail

206.
=   *   *     0  0     sfo      sfofa fofsfo      8   5   0   1   0   3 sofa

207.
=   *   =     0  0     fru      frumu fumfrufu'u  6   3   0   0   0   3 frown

208.
=   *   *     0  0     gra      grake gakgra     20   2   0   1   0  18 gram

209.
*   *   *     0  0 gic gli      glico gicgli     42  35   0  29   0   7 Englis
not changed; sacred

210.
*   *   *     0  0 jaf jma      jamfu jafjma     53  44   8  22   8   9 foot
not changed; mildly sacred by heavy usage, if not by anyone's particular
memory - people regularly have chosen jaf over jma for some reason, even
when it forced hyphenation.  The best example that overkill may be an
intellectual aesthetic.

2    badjafyde'i
2    badjafyde'icalku
2    jaftamji
2    jaftamjycalku
10   jafti'e

211
=   *   *     0  0     kle lei  klesi lesklelei  63  24   1  14   3  39 class
changed but ambivalent on this; lei retained since it matches cmavo; may
want to keep 'les', too in order to have a good second instance of all 3
rafsi besides condi - one that will minimally hyphenate with condi
rafsi.

212
*   *   *     0  0 sev sne      senva sevsne     27  19   2  12   0   8 dream
not changed; significant use of sev over sne

2    sevdrari'a
1    sevycinse
10   sevycu'u

213.
=   *   *     0  0     smo      smoka soksmo      5   3   0   0   0   2 sock

214.
=   *   *     0  0     sfu      sufti sufsfu      0   0   0   0   0   0 hoof

215.
*   *   *     0  0 val vla      valsi valvla    154  73  19  34  12  81 word
not changed; significant use of val over vla

1    valcku
1    valdre
4    valgerna
4    valkei
2    valnunkei
6    valselkei
1    valste

216.
=   *   *     0  0     vre      vreta vetvre     23  18   4   6   1   5 reclin
changed but ???; some use of vet over vre, which remains unused

1    vetspi

217.
*   *   =     0  0 vor vro      vorme vorvrovo'e 38  11   0   0   4  27 door

218.
=   *   *     0  0     xru      xruti xutxru     36  16   0  10   0  20 return

219.
=   *   *     0  0     zma mau  zmadu zadzmamau 177  62  16  26  12 115 more
mau sacred, even zad has seen usage, but zma does predominate

4    xagzadri'a

1    geizma
2    kazmaksi
2    zmadji
1    zmagei
14   zmanei
1    zmanelci
9    zmapluka
2    zmari'a
1    zmasatci
2    zmaslarai
2    zmaxau

5    badmau
2    bitmau
3    clanymau
2    cnimau
1    dubjavmau
1    galmau
1    glemau
1    jbimau
4    ladmau
1    lirmau
1    maugle
1    maurderxi
6    maurzau
1    nalxagmau
3    pa'ermau
3    racmau
2    rinxagmau
1    selkemselgeimau
1    selkemymaula'e
1    selselgeimau
1    selvlimau
2    selyla'emau
3    tagmau
4    tecmau
2    tepri'amau
2    tervlimau
4    tilmau
2    tormau
63   vlimau
31   xagmau
3    xagmaubi'o
1    xagmaugau

220.
=   *   *     0  0     zmi      zmiku zikzmi     16  12   4   7   0   4 automa

221.
=   *   *     0  0     zdi      zdile zilzdi     44  32   3  16   6  12 amusin
changed, but ???; some usage of zil over zdi, but not a lot

6    zilsri

1    drizdi
2    kelzdi
1    nunselzdi
6    nunzdi
4    selzdi
2    selzdigei
10   zdifanza

222.
=   *   *     0  0     zmu      zumri zumzmu      7   7   1   2   0   0 maize