WikiDiscuss

---

> Chiming in with a rare re fepni:
>
> John E Clifford wrote:
> >>It is not clear what "being in the language"
> >>means to you.
> >>All BAIs under discussion have been listed in
> >>the ma'oste
> >>for years. None has been added since I joined
> >>Lojban
> >>in '94.
> >
> > How many of them have been used?
>
> Not a valid argument. I wouldn't be surprised
> if more than half the
> cmavo have never been used, in part because the
> few people doing most of
> the writing over the years thought that the
> definitions in the cmavo
> list were inadequate.

An argument for either editing the list or, at
best, shifting to the "No usage, but a probable
meaning somewhere in the area of ..." This will
work less well with things that are not — as BAI
-- vaguely linked to something with an
established meaning, but still somethin useful
can usually be said, without binding a future
user to that if he needs a new usage.

> A brief history of BAI
>
> Recognizing that JCB had a bunch of different
> things - including case
> tags, causals, and modals, we decided to treat
> them all as one
> interchangeable selma'o now called BAI. The
> causals and modals among
> them had a history of being convertible using
> SE, TE, etc, and the
> causals of being negated (leading to the
> because/therefore/despite/nevertheless set of
> causals).
>
> Because they were grammatically convertible, we
> looked at what
> conversion and negation would do for other BAI
> members. I believe this
> was considered at the time of the negation
> research you did. I noticed
> that several BAIs could be used in this way,
> especially the one based on
> jalge, which was a kind of backwards causal.
>
> People complained (this might have been Nick,
> and Cowan too) that while
> each BAI was nominally associated with some
> selbri and picked out one
> place of that selbri, it was unpredictable
> which place was the one of
> most interest. Some BAIs had several places
> useful while others seemed
> to be either less useful or at least not
> corresponding to some English
> equivalent.
>
> It was then proposed that all BAIs be oriented
> in the same way as the
> selbri they were derived from and take the
> meaning rigidly from that
> selbri. We decided that in <BAI sumti>, the
> sumti had the semantics
> of x1 of the source selbri, SE BAI the x2,
> etc., making the memorization
> easier.
>
> Since we had a pretty fixed set of selbri, I
> attempted to figure out the
> plausible meanings of all the conversions of
> all the BAIs based on this
> rigid association of selbri. I believe that
> Cowan and Nora reviewed
> this list. We recognized that some of these
> conversions would be
> relatively useless, but we weren't about to
> constrain things because
> right about that time, the evolution was
> tending the other way, and
> people were using FIhO+gismu to create all
> sorts of ad-hoc tags.
>
> We baselined the cmavo list in 1994 with all of
> the conversion compounds
> listed.

In other words, this list was indeed generated
mechanically, with minimal regard for possible
meaning (indeed, despite evidence that some
proposals were junk). That was what it looked
like but I am still sorry to hear that it was so.
Given that history and the fact that we have made
it difficult to do anything more about it than a
shift from "definitions" to suggestions for
potential use, I suggest we take the latter
aproach.

> > That is the
> > most significant feature of being in the
> > language.
>
> Since "usage" has been a matter of significance
> for only a small number
> of people until shortly before byfy got
> started, I have contended all
> along that there could not possibly have been
> enough usage to decide
> whether something is "useful" or not.

Hey, this is a dictionary being constructed here,
not a word list laid down from on high
(presumably), so usage always counts. But, since
it inherits some of an unfortunate word list, the
proper thing to do is include the unused items
with a note about the area of their use but
without nailiing that usage down so as to
preclude other less direct uses as the first to
come forth.

> > So, given a BAI
> > in the language as a BAI, any SEBAI could be
> in
> > the language as a BAI. At some point — I
> don't
> > much care when, though it was after whenever
> my
> > paper list was made up — a number of BAI
> were
> > added on the basis — as far as I can tell --
> of
> > that second sense of being in the language.
> It
> > may be that some few of these have been used
> (my
> > list has a large handful of SEBAI, enough to
> > establish the pattern). The rest seem merely
> to
> > be an excresence.
>
> For each of the BAI in the language, there is
> at least one place that we
> could see a need for. The regularizing of BAI
> semantics means that all
> the other options are permissible, but it is
> precisely because there has
> NOT been a good definition for these that we
> cannot judge based on
> usage. We need a 5 year period of usage AFTER
> we have good definitions.
>
This is bass ackwards in the extreme: you can't
have a good definition until you have usage.
Witness what is happening even with the
prescriptive definitions of words that are used:
the meanings are sliding to fit needs (at least
in the direction of clarity if not farther).
This is one of the reasons why BPFK esists after
all.

> >>>but my protest is exqactly to including them
> in
> >>>the dictionary when they have never
> occrredin
> >>>the language. Will all the 100000 or so
> gismu be
> >>>included?
> >>
> >>Not sure what you mean by that. There are
> >>1400 or so gismu in the language.
> >
> > But around 100,000 CVCCV and CCVCV forms.
>
> The list of gismu, just as the list of BAI, has
> been baselined for over
> 10 years.

Relevance? The point was that if we are going to
list all of the permutations on BAI allowed by
some rule and then dream up meanings for them,
why not do so with other forms generated by some
rule: all the potential gismu or long-form or XVV
cmavo or... ? It was meant to be a reduction to
absurdity, to which the reply "But there is a
closed list of gismu" does not answer.

> It takes an act of byfy to delete something
> from the baseline just as it
> would take such an act to add to the baseline.
>
> I oppose any argument for deletion based on
> usage when usage is
> dependent on having good definitions.

This is again just wrong way round. This
implies, for example, that the only lujvo that
can be used are those that have already been
defined, for only those have definitions. But
their definitions (usually) came from someone
using them, not use after definition.

>
> >>>What about all the cmavo in /x/?
> >>
> >>I would include {xa'o}, but I doubt I'll get
> my
> >>way.
> >
> > Butwhat about all the rest? This is a
> > particularly interesting case, because it
> brings
> > in forms whose meaning cannot even be guessed
> at
> > in advance, while I suppose there are some
> limits
> > (though I would hate to try to figure out
> what
> > they are) to what a SEBAI might mean, given
> the
> > BAI and perhaps the "related brivla."
>
> The XVV cmavo were by the baseline
> experimental. It is permissible for
> the byfy to change the baseline and assign
> those XVV to specific things
> that have been experimented with, but I would
> oppose doing that until
> all of the existing things have been defined.
> One option proposed from
> the beginning would be, when there is
> controversy between two different
> semantic interpretations of some cmavo, that
> the least controversial
> solution would be to split the cmavo into two.
> That was specifically
> considered during the gadri debate.

See above.

(Snip irrelevant harangue about XVV etc.)

>>>But unfortunately the BAIs are not in
>>>experimental
>>>space. They are a standard part of the
official
>>>cmavo,
>>>all of which need defining.
>>
>> No, they need editing if they are already in,
>>and
>> a note about how to get them back in if a need
>> for them arises,

>That is not the sort of decision that the byfy
>should be making yet.

Well, OK; I have suggested an alternative
(indeed, one I always had in mind — even in
words, I think).

>> Actually, we seem to me to agree rather well:
>> these things do not at present have any
>business
>> being presented as a part of actual Lojban,
>> worthy of a line in a dictionary.

>They are part of the baselined language. It
>takes a formal vote on each
>one to delete it, and I would protest such votes
>taking place in the
>middle of the quite different task that the byfy
>is now engaged in of
>trying to define the words.

See above.

>For BAI, we should not need actual usage to
>define what the conversions
>mean - that was precisely the argument that was
>made for regularizing
>them. We need the examples before there will be
>usage.

This presupposes a connection between BAI and
gismu which is not justified by the actual cases
nor argued for in any place I can find (and is
explicitly denied a couple of times). If it was
in peoples' minds when this expansion project was
carried out, it did not guide them well in cases
of actual use and led to absurdities in the
phantasmic cases.

>And in the case of BAI, I think usage patterns
>here will be strongly
>correlated with the native language of the
>speaker, so deleting things
>based on nonusage when the language is dominated
>by English speakers and
>one Spanish speaker strikes me as asking for
>bias.

An interesting notion, which should be looked at.
It is true that native usage does affect much of
Lojban usage, so it may well here. I don't see
any evidence of it, but then, it's my native
language that is being imitated.

>>>Well, that's the kind of clarification I'm
>>>after. Should
>>>{du'o} be {fi'o djuno}, or something much
>>>wider?

>The baseline definition bases it on djuno. If
>someone wants to propose
>a lujvo compound of djuno I could probably
accept >it, but I think we
>need BAI for all of the djuno places.

Usage — indeed even the official defintion --
separates it from {djuno} and ties it with
something more like {jinvi}; theusage with
{djuno} would be appreciably less useful.

>>>>and {ri'anai},
>>>>whose meaning violates both {rinka} and
>>>{nai}, as
>>>
>>>>typically understood in this game of BAI
>>>creation, and is not easily constructed in
>>>this
>>>
>>>>way from other brivla.

>It seems plausible without checking that the
>causals may currently not
>be aligned with the other BAIs. They were
>included in the language
>based on JCB, and merged with the others before
>people decided that they
>wanted systematic meanings, and I may have
>required retention of the
>old meaning for those as a condition for
>systematizing the others.
>Since I suspect that the causals have NOT been
>used as much as I
>expected based on the importance that they were
>in the TLI era, I would
>accept a general realignment of the causals so
>that they match the other
>BAIs.

Actually, since the causals by and large make
sense and many of the others do not, I would
expect that careful usage would shift things the
other way. But since we are now allowing that
usage will affect definitions (jumping from the
17th to the 2oth century in two paragraphs), why
not let it go all the way and leave the lot of
those not yet used to be decided by usage (within
broad limits of course — I see no way of getting
rid of the tendency to associate BAI with gismu
in some way).

>However, if this means that we do NOT have the
>foursomes of
>because/therefore/despite/nevertheless, then I
>will insist that a
>solution be found for representing all 4 of
those >for each of the
>causal-like BAI members.

I can't check because it is hard to figure out
what keywords muight be used for some of these,
but it seems likely that all are there, since
they were already in Loglan.

>> But the official line does not contain {to'e},
>> nor ought it Zipfily. And the claim that it
is
>> regular presupposes that there are rules,
which
>> is not obviously the case.

>The causals had rules, but they may not be
>consistent with the evolved
>gismu that they are based on (some of which were
>changed due to
>sumti-raising concerns), and changes to negation
>might also render them
>less fitting to the pattern.
This being the case, it seems a bad idea to have
built a mass of other examples on the rules from
them-- which rules they do not themselves follow.
But the point remains that the ruled definitions
destroy the possibility for using BAI in useful
ways if the useful way does not happen to fit the
rule. To be sure, the likelihood is that the
ruled definition will eventually be ignored if
the usage is useful, but why set up the
roadblocks to begin with.

>I won't suggest the solution since I have not
>followed the debate and
>can't without checking know the currently
favored >ways of negating a BAI
>and the semantic import thereof (i.e. na'ebai vs
>bainai vs na bai). You
>pc should be aware more than most of the history
>of JCBs causals that
>led to the semantic foursome being associated
>with conversion and
>negation of each causal. Just make sure that
the >result remains
>consistent, and capable of expressing all the
>stuff that JCB had in mind.

The causal cases are set up and used, so are not
the problem (unless some of them got lost in the
"systematic" rewriting); the problem is forcing
these patterns on everything else, whether they
produce useful results or not and regardless of
whether they bury useful meanings or not. It may
be, of course, that every one of the so far
unused expressions that ever does get used is
used for exactly what the system predicted and is
so used not because of prediction but because of
the internal logic of the usage, but that still
does not make the predictions justified
prescriptions.