Naturalness, etc. (was: Re: LUNATIC again)

On Wed, 4 Nov 1998, Logical Language Group wrote:

> I think that merely requiring language to be "natural" makes point 3
> operative, as you have defined natural.  Languages with only one (remaining)
> native speaker are studied for preservation purposes, but are considered
> essentially dead, and certainly not subject to study for their
> natural evolution, because it deosn't much happen at that point.  Languages
> which do not yet have significant numnbers of native and/or fluent speakers
> don't undergo anything like natural evolution.
Sorry, I wasn't being clear.  First of all, I and the linguists I was
referring to do not *require* a language to be natural.  According to my
usage, naturalness is not part of the definition of "language", but
rather part of the definition of "natural human language", a narrower
concept.  If we construe "natural human languages" (the subject which
linguists study) as a proper subset of "languages", then we can allow
the latter term to include other things like computer languages,
pidgins, scientific notation, etc..

Also, with regard to "natural":  What I should have said is that "natural"
refers to "naturally evolved" rather than "naturally evolving".  So
languages which are dying out, and are thus not currently doing much
evolving, might still count as "natural", if they originated as a result
of natural evolution.  I'm not so sure about constructed languages which
are intended to evolve naturally but haven't yet begun to (like Lojban).
My suspicion is that many linguists would exclude these from the definition
of "natural human language" on the grounds that they did not *arise* (in
their current form) by natural means.  (Cf. pidgins and creoles.  Pidgins
are not "natural", but creoles are, according to this definition.)

In my opinion, though, even if Lojban does not count as a "natural human
language", it still counts as a "language".  In defining "language" in
this broader sense, I consider the following criteria to be essential:

(1) Arbitrariness:  Concepts are associated to symbols, where this
association is conventional (standardised) and largely non-iconic.

(2) Compositionality:  Expressions of the language are composed of
smaller expressions belonging to the same language, and can in turn be
combined into larger expressions.  Expressions are formed according to
a finite set of rules.  An expression is interpreted by knowing the
meanings of its parts and how those parts are combined.

(3) Creativity:  Given finite means (a finite vocabulary and a finite
number of rules for forming expressions) it is possible to produce an
infinite number of utterances.

As far as I can see from having read the Lojban Reference Grammar,
Lojban fits all three of these criteria.

> >Unfortunately, many formalists - and I say
> >this as a formalist myself, a Chomskyan even (ooh! *shudder*) - do not
> >make use of actual examples of spontaneous language use, but base their
> >theories entirely around elicited data and speaker judgements.  Myself, I
> >try to use both kinds of evidence, respecting the primacy of spontaneous
> >textual material, while using elicited data to confirm or discount my
> >hypotheses.  Nevertheless, I adopt the Chomskyan premise that language
> >is best viewed as a system of knowledge rather than a system of
> >behaviour (insofar as they can be separated), and thus I would not
> >include your point 3 in a formal definition of "language", although I
> >might use it as part of an informal characterisation of language.
>
> A polite Chomskyan! What a pleasure %^).
Are we really so rare as to be a pleasure?  :-)  In my experience it's
the functionalists who are the rude ones, always referring to us formalists
as if Generative Grammar were some sort of religion with Chomsky as supreme
and infallible pontiff!  :-)

> >Here I find the term "model language" useful and appropriate.  A model
> >language is and is not a language in the same way that a model airplane
> >is and is not an airplane.
>
> I use that term also, and would live to see it adiopted by conlangers in
> place of the other term.
Gotta be shorter, though.  Modlang?  Sounds too much like "modern language"
or something...

> And Lojban differs from this in that we hope that, by having a significant
> user community, that the language will not only be used for something in the
> manner of a "real language" that it will acquire the complexity and completeness
> of such a "real language" in relatively short order, while preserving its
> essential core in some way recognizable from the prescription (and presumably
> offering maximal chance to actually TEST some of the Chomskyan andbehaviorist
> theories of language by starting with something rather unlike the typical
> natural human language and seeing if after undergoing natural human language
> evolution it shows characteristics predicted by the various schools of
> language.
Here's a question I've been curious about ever since I heard about
this proposed experiment:  Supposing the experiment ever gets carried out,
how would you control for the influence of natlangs on your Lojban-speaking
population, since presumably all of them will be bilingual?  In other
words, suppose that, after undergoing natural human language evolution,
Lojban loses all of its distinctive features and comes to closely
resemble a 'typical' natlang (to use your word). How would you know what
aspects of that change were due to speakers adapting Lojban to suit the
innate requirements of their 'mental grammars', and what aspects were due
to direct contact with natlangs?  It's well known, after all, that in
situations of widespread bilingualism, languages can evolve to resemble
each other more closely, not only in vocabulary but in syntactic and
semantic structure (witness the 'Spanglish' widely spoken here in LA).
I don't imagine you propose to isolate your community of Lojban speakers
from the rest of the population, so...

Matt.