baselines and semantics

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

And writes:
>What I find particularly odd is Lojban's conjunction of rigid baselines
>on the one hand, and on the other hand leaving vast tracts of the
>language to be determined by usage.  For example, gismu place structures
>are baselined, but the meaning of the gismu is merely vaguely indicated
>by a keyword, and the Lojban community is left to negotiate the word
>meanings among themselves.

>We are told that Lojban users complain if the grammar changes, yet most
>of it - the semantics - never will be baselined.

>From what Lojbab has said to me, it seems that the policy is to
>baseline as much as possible, except for the semantics, because Lojbab &
>his colleagues haven't the time or (he claimed) competence to create the
>semantics.

Furthermore, I'll claim that NO ONE has the competence to define a
prescriptive semantics, because there is no unambiguous mode of
expression to communicate semantics.  Hence we can only through
induction come up with a model of semantics.  I believe that that model,
if created, would differ for every human being, since semantics is a
function of personal idiolect, not of a language as a whole.

(Translated to English, that means "words mean what I want them to mean"
with "I" moving with each new speaker of the phrase.)

There are some items of the semantics of the language that can be
defined as part of defining each word, but that definition is untested
by usage to see if it works and covers all cases.  I won't pretend to be
omniscient enough to cover all possible expressions of meaning.  It is
an untested assumption basic to the language that the claims of
logicians to the effect that a predicate structure can handle all such
expressions - therefore an unambiguous predicate structure that has
unambiguous bells and whistles to simplify the most obvious usages of
natural language is possible based on that assumption.

>This vacuum where much of the semantics would be in a natural language
>undermines the merits of the prescriptiveness of the rest of the grammar.

The merits of prescriptiveness in Lojban phonology and grammar is that
the language is unambiguous in those areas.  I don't believe a
completely unambiguous semantics is possible, and I don't believe that
most Lojbanists want it - hence the ever-recurring concern with the
ability to joke, write poetry, or express metonymy in the language.  We
don't yet even know for sure that the prescribed rather loose Lojban
handling of sumti-raising can be taught because it is not a conscious
process for most language speakers.  Yet unambiguous sumti-raising would
be a vital first step in a semantic prescription.

>It would make more sense to:
>(a) abandon all baselines, simply offering offering the fruits of the
>Lojbanists' considerable linguafactive talents to the community to do
>with what they will,

In effect, that is what we plan to do, after using the prescriptive
baseline as a crutch to give people confidence in stability - that the
language will not change like sand beneath their feet to such a degree
that they will have to relearn things with every new design iteration.
The concept of baselines says that a later design change must not change
a previously baselined feature for reasons short of being severely
broken.  Areas baselined early are areas that people learning the
language have been burned before by having them changed underfoot.  Even
tough we do occasionally make a baseline change, it almost invariably
has been in an area not yet written up fully, and not significantly
learned yet, and even then is psychological recognized as an addition,
rather than a change.

This can't be implemented prior to the language being done, and having a
large enough speaker base that we have an established norm that has some
psychological manifestation rather than a paper one, such that we are no
longer reliant on 1 or 2 (all too human and error prone) people to set
the examples.  At that point, the norm itself replaces the baseline.

>(b) make a commitment to the eventual baselining of all the grammar,
>including the semantics (and add a couple of decades or more to the
>date for the final baseline).

This inherently defines why we won't do it.  I don't intend to spend
that kind of time completing the design of Loglan.  Even if I wanted to,
I can't see the community (or Nora for my personal work) supporting a
language they can't yet learn and use.  I have spent longer than I
intended as it is.  Nor are there others who want the job to my
knowledge - if there are, they of course have the option to pick up the
language where we stop prescribing and spend those decades.  I certainly
won't stop them; that is why the language baseline is in the public
domain - so I CAN'T.  But people will be learning the language faster
than it can be prescribed, so I view such an activity as a waste of time
other than as a theoretical effort.

The bulk of our prescription is being made to enable us to teach the
language at a distance.  I am not a prescriptivist at heart, only as a
teacher (and I don;t claim to be a good one).  Our emphasis must turn
from inventing a language to teaching it.  Colin and Nick and Mark, etc.
are demonstrating with each new writing effort that there is something
there we can call a language and that at least some people can learn it.

So let's be done with it.

lojbab