LINGUIST List 7.1490

Wed Oct 23 1996

Disc: Natural language

Editor for this issue: Susan Robinson <robinsonemunix.emich.edu>

Directory

1. Stirling Newberry, Re: 7.1480, Disc: Natural language
2. benji wald, natural language

Message 1: Re: 7.1480, Disc: Natural language

>Date: Thu, 17 Oct 1996 14:42:52 -0800
>From: hendrikuvic.ca
>Subject: Re: 7.1463, Disc: Natural language
>
>Having read many of the interesting contributions on how to discern
>natural languages, i feel reminded very much of a few issues from
>Meathematics:
>
>If i have understood the matter correctly, it has been established (by
>Goedel et al) that in any given open system, i.e., one that contains a
>potentially infinite number of elements and operations, there may
>occur statements that are "true" but unprovably so and also some that
>may be unprovably "false" - much to the disappointment (in both ways)
>of those mathematicians who until then believed in the existence of an
>algorithm - searched for but not yet discovered - for distinguishing
>with certainty "true" from "false" statements.

To be precise any system based on bianary (true/false) logic cannot be
both complete (all true statements provable) and consistent (all
statements provable are true).

It can also be shown that it can contain at most an alpha-null number
of predicates.

>Another issue: bivalent or Aristotelian logic is nowadays being seen
>(at least by some) as one of many possible logic system - quite like
>Euclidean geometry, once perceived to be the only possible geometry,
>was relativised by Riemann. ("Fuzzy logic" is the term for
>incorporating this new understanding into statisitcs and computer
>applications.) This does not mean Euklidean geometry is suddenly
>"wrong"; it is now perceived as a subset instead of the whole set and
>can thus only be applied correctly in the appropriate context.

This can be shown to be the case - the value q is the superposition of
true and false. q based systems can avoid the Goedel limit and be both
complete and consistent and can contain an number of the continuum
number of predicates.





Stirling Newberry
Boston, Massachusetts
allegrobluesky.net
newberryopenmarket.com
Starting October 15th:
http://www.bluesky.net/~allegro

I've only caught the last few exchanges on the topic of "natural
language" on the list. so I run the risk of making an off-the-mark
comment. However, my attention was seized by the assumed dichomtomy
between "natural" and "artificial" lg, where the latter is exemplified
by such things as Esperanto and other consciously constructed
"international auxiliary languages". My thought is that, as far as I
can see, there is nothing in the consciously constructed rules for,
say, Esperanto, that is alien to natural languages. Therefore, I
would expect that the hypothetical Martian linguist would not
recognise it as anything but a natural language (though being astute
s/he might wonder why it doesn't have any of those cute
"irregularities" that we love so much, and no doubt Martian linguists
love too -- and might be the tip-off that the language is "new",
cf. creoles?)

The reason for the naturalness (I aver) is that the inventors of such
languages were using their "innate language faculty" to create them,
dragging that baggage into the creative process. A rough analogy
might be that many people, children no less tha others, are capable of
composing new songs appropriate to their cultures, but are not
consciously aware of the structure of those songs, say, the ABA 12-bar
structure, or whatever. They just internalise it from exposure. Of
course, there are limits to this analogy, focussing mainly on the
unconscious reception of structure. "Innate faculty" assumes more
deeply that external exposure is not the ultimate source of the stuff
that is built into the artifical language. Natural languages are
famous for the vast majority of their rules being inexplicit, i.e.,
never rising to awareness (except among Martian linguists). This is
no doubt true of Esperanto and the like, though such rules are
invisible in pedagogical grammars of such languages (as of NLs).
Similarly, it does not have ridiculous rules like "make every second
word a noun" or "alternate between sentences which have the square and
the square root of the morphemes of the last sentence", "alternate
between prepositions and postpositions of the same form", and so on.
These are the kind of rules that Chomsky long ago observed are alien
to natural language (and "universal grammar"). They do not occur in
any artificial language, which is indeed only superficially different
from a natural language, if we take Esperanto as a typical example.

For those who felt or asserted that the difference between Esperanto
as originally invented and as spoken now somehow reflects the
difference between a "natural" and "unnatural" language, I am curious
about just what differences they have in mind. (uhm, second languages
are "natural languages" too, or am I wrong about this?)

In sum, it seems to me that Esperanto and other creations of that type
are nothing more than parasitic creations based on natural language/s.
In saying this, I may not be aware enough of some of the fine points
of consciously created Esperanto grammar. Are there any rules that
are not found in natural languages -- or, maybe more difficult to say,
"impossible in NLs"? == Benji (Yeah, I got a new e-mail account. Hi).