LINGUIST List 7.1430

Sun Oct 13 1996

Disc: Natural language

Editor for this issue: Ann Dizdar <dizdartam2000.tamu.edu>

Directory

1. "Ellen F. Prince", Re: 7.1405, Disc: Natural language
2. Jouko Lindstedt, Disc: Natural language

Message 1: Re: 7.1405, Disc: Natural language

i've just read this thread for the first time and am surprised that
it's controversial. fwiw, i've always understood _natural language_
basically in larry trask's way -- a language that can be acquired as a
native language by humans. of course, since we cannot yet characterize
such a language, we must stick with those languages that actually were
or are someone's native language. in fact, for many many linguists,
the goal of linguistics is precisely to arrive at such a
characterization.

btw, esperanto is such a language -- i have met children with
esperanto as their mother tongue. of course, how it differs from
zamenhof's creation is an interesting research topic still waiting to
be studied, so far as i know.

I have received lots of thought-provoking reactions to my posting
about the notion "natural language". First I want to thank Ming-wei
Ernest Lee, who pointed out John Lyons's excellent discussion of the
different meanings of "natural", included in his book _Natural
language and universal grammar_ (=Essays in Linguistic Theory, Vol. 1,
Cambridge UP 1991). I was a bit ashamed for not knowing that source,
but neither did any other colleague make a reference to it. I have
only just acquired the book, so I cannot sketch my "final" opinion on
the issue yet -- perhaps I should write a paper -- but some points in
the messages and postings I have read are perhaps worth commenting
already.

Several people noted that the term "natural language" is most often
used to contrast human languages with computer programming
languages. Some think that the usage has actually originated among
computer scientists, but Martin Haspelmath suggests a broader
background (personal communication):

> I can offer no specific evidence for this, but I think that
> the term "natural language" was coined by philosophers (who
> were dealing with all sorts of artificial languages, from
> Leibniz's lingua mentalis to programming languages) to
> refer specifically to "natural" human language, i.e.
> excluding Esperanto is not the primary goal. Of course, we
> would say that "artificial languages" are "languages" only
> in a metaphorical sense, but on the other hand they are
> clearly human.
> Anyway, given the philosophical background of the term, we
> should avoid it simply for that reason: As linguists, we
> shouldn't call our object of study by a term that is mainly
> useful to philosophers, logicians, etc.

As Becky Moreton (p.c.) noted, if you see "natural language" in 
the name of a conference announcement, it is most often a 
computer-oriented conference (on "natural language dialogue 
systems" or the like). In a similar vein, in The Encyclopedia of 
Language and Linguistics 1-8 (Ed. by R. E. Asher & J. M. Y. 
Simpson. Pergamon Press 1994) all the articles whose titles begin 
with the words "natural language" are about computational 
linguistics. In the key articles about the properties of language, 
the terms "human language" or simply "language" are used much more 
often than "natural language".

Peter Daniels wrote:

> "Natural" is simply a convenient antonym of "artificial"
> (cf. "natural foods" which contain "no artificial
> ingredients"). Thus "natural language" is the unmarked term
> in a privative opposition and requires no particular
> definition. Also, it's an ordinary-language term, not a
> technical term, and can be used intuitively.

I have seldom seen "natural language" used outside linguistic
discourse, and at any rate, I was asking for a definition, not for
intuitive use. Defining "natural" as "not artificial" won't work,
because "artificial" lumps together most various entities. Turbo
Pascal, the predicate calculus, Esperanto (with a speech community
including native speakers), Jespersen's Novial (with almost no
speakers of any sort), Old Church Slavonic, the new Bosnian standard
language, and (as suggested by Larry Trask) high Basque used by
television newsreaders have all been called "artificial" by serious
linguists at some time or another. So it rather seems that
"artificial" is only negatively defined, "natural" needs a positive
definition. To distinguish Esperanto from programming languages, it
has been called a "constructed", "planned", "invented", "contrived",
"synthetic" or "artificial natural" (!) language. This is a
terminological mess.

Herb Stahlke (?) mentioned Charles Hockett's "design features of human
language" (NB., not "natural language"), and I think this is a
fruitful approach. Looking at the list of the sixteen design features
(as presented in the Encyclopedia mentioned above), it is easy to see
that computer languages do not fit them. As for Esperanto, when it was
created it already fitted them all, though one (learnability as the
first language) was proven only later. But if I understood Lyons's
point right (upon the first cursory reading), it is quite trivial that
a language derived from natural languages and modelled on them behaves
structurally as a natural language. The question is, how much language
standardization and planning may change the "basic" language(s) before
we get something "artificial". For Larry Trask, some standard
languages are articial (or at least more artificial than others); and
this is something which always sounds value-laden, even if it is not
meant that way. "Language engineering" sounds bad, too; and what makes
a language "tortured"?

The big problem with Hockett's features is that they were designed
with only spoken, not signed languges in mind. I cannot help seeing in
this a late reflex of Edward Sapir's total ignorance of the true
nature of sign languages (in the introductory chapter of _Language_ he
characterizes all sign languages as "transfers" of spoken
languages). And most lists of "universals of human language" are still
made with the tacit assumption that sign languages are excluded, as
Esa Itkonen, among others, has pointed out.

Joseph Foster presents a kind of operative definition:

> There is however, or so seems it to me, a set of
> differences between a language like Estonian or English and
> one like Esperanto. For instance, are the stops of
> Esperanto aspirate or nonaspirate. How can we find out?
> Dr. Zammenhof didnt tell us--presumably he was unaware
> languages could differ in this way--and there are no
> "uncontaminated" native speakers who can serve as final
> authroities.

I assume that Foster has not read all of Zamenhof's writings, and
neither have I, so we cannot discuss whether he was aware of
aspiration (it would be odd if he wasn't). But of course it would be
completely beside the point to look for the characterization of the
Esperanto stops in the writings of somebody who died in 1917. You
should go to places where Esperanto is spoken now and find out. This
may first seem useless: as most Esperanto speakers have interference
from their native languages, some have aspiration and some do not. But
notice first that there are languages (such as Swedish) where some
regional varieties require aspiration, some exclude it (as the Swedish
of the native speakers in Finland), so this would certainly not
qualify a language "unnatural" in structural terms. But what is more
important, there is a vast agreement among Esperanto speakers as to
whose pronunciation "sounds best"; on this basis it is feasible to say
that Esperanto stops do not have aspiration, for even those speakers
who aspirate would like not to do so. The pronunciation norm of
Esperanto has never been really codified or even described
extensively, but it exists as an implicit norm in the community.

As for the criterion of "uncontaminated" native speakers, I assume
that this is a reference to the lack of monolingual native speakers of
Esperanto (all are bilingual or trilingual). But it is easy to find
lots of minority languages (Sami, also known as Lappish, for
instance), whose adult speakers are practically all bilingual. I
suspect this criterion would in fact exclude the majority of the
world's 6000 languages, so I don't think it can be seriously
maintained.

Although in some messages that I got the criterion of spontaneous
origins was considered essential (to the point of excluding Modern
Hebrew), I think many would find Larry Trask's "mother tongue"
criterion more fruitful. He wrote:

> American Sign Language is a natural language for the same
> reason, at least as used by native signers. So is Israeli
> Hebrew. I can see not the slightest difficulty here in
> reaching this last decision; what is problematic about
> Israeli Hebrew is the nature of its relation to Biblical
> Hebrew, but this question is neither here nor there in the
> present context. And, by my criterion, the Esperanto
> spoken by native speakers (if these exist) must also be a 
> natural language. The non-native Esperanto of other people
> is a different matter, but then, when I declare that
> Spanish is a natural language, I certainly don't have in
> mind my own halting Spanish.

This would mean that the big change for Tok Pisin was when it got
native speakers; but I have understood the issue is not yet settled
whether the crucial step was indeed creolization, or rather the
increasing use in different functions as an "extended pidgin" of a
speech community. Similarly, I cannot distinguish an adult native
speaker of Esperanto from any other fluent speaker of that
language. Perhaps if there were a community of native speakers, it
would create peculiarities of its own; but this means that the
community is more important than the "mother-tongue- ness". And as
Nancy Frishberg and others assure me, deaf people can quickly become
fluent in a sign languge even if they are not exposed to it at home
and at an early age, as is the case with typical "mother
tongues". This is not meant to suggest that it is not the native
language for them, but to emphasize that for fluency you necessarily
need contact with the community, not necessarily (though preferably)
an exposure in early childhood. So, although I agree with Trask about
the importance of the "mother tongue" and "speech community" criteria,
I would weigh them in a different fashion.

Finally, most of the time we only speak about "language" and 
"languages", and it must at least be thought twice to see whether 
"natural" really adds something we can define. I'll go back to 
reading Lyons. Thank you all for stimulating ideas!

Jouko Lindstedt <jouko.lindstedthelsinki.fi>