Cladistic language concepts

Tue Sep 8 13:22:20 UTC 1998

----------------------------Original message----------------------------
    I have really enjoyed Michael Ghiselin's contribution to this list.
Analogies from biology have been very powerful in linguistics (as many
of the responses have shown.) So it's comforting to see that so many of
the foundational concepts in biology are as problematic and ill-defined
as the foundational concepts in linguistics. It is also nice to have
someone outside the field, particularly a biologist, raising the
important issue of the ontology of language.

    As I see  it, linguistics since its inception in the nineteenth
century, has been defined not so much by a set of problems or a class of
phenomena to be explained, or even by a methodological approach, but
rather by the strange motto "Linguistics must be made a (natural)
science!" That is, to begin with, there was a desire to break with the
tradition of studying language within the medieval curriculum of
grammar, logic, and rhetoric. Two of the most influential research
programs in linguistics, the Neo-grammarian in the late nineteenth
century and Generative Grammar in the mid-twentieth embraced this motto
by asserting strong connections between linguistics and biology. The
neo-grammarian family tree model asserts, in effect if not explicitly,
that language is an organism independent of its speakers.  The
generative model asserts that language is a part of an organism, an
organ, like the hand or the liver.

    Beside the fact that these two models are logically incompatible (if
the first model is acurrate, then language change must be due to
mutation and speciation motivated by selectional pressures, if the
second correct then the locus for this change must be the human organism
itself-- hence language must be due to mutation and speciation within
the human race, which is empirically false), attempts to impose
biological models on language study have had the paradoxical effect of
making linguistics less empirical, hence less 'scientific'-- by a priori
limiting the types of data deemed worthy of consideration as well as the
types of hypotheses which might be proposed.

    Thus the traditional position  in historical linguistics is that
'genetic' relationships between languages are somehow real and
important, while relationships due to contact is mere noise. It has
taken a long time for linguists to accept the idea that contact
relationships are also subject to regularities which can be rigorously
modeled.  But really there are no genetic relationships. All
similarities between languages are based on contact (since an infant's
first language is the language of those who raise it, not necessarily
its biological parents), and so-called 'genetic' relationships simply
represent one extreme on a continuum of contact.

    The faults of generativists in this regard are even more egregious.
At one dark point it was widely asserted that it was a waste of time for
linguistics to study languages, since universals of language could be
directly intuited through introspection of one's own native language.
(Sort of like, one doesn't have to dissect thousands of cadavers to
understand the principles of anatomy-- one will do). Generalizations
proposed on the basis of data from a variety of languages could be
dismissed as mere facts about langauges which happen to be statistically
true-- as opposed to insights about the 'language faculty' which were
necessarily true. This has all changed now. Generative grammar in its
current form is largely a set of reactions to discoveries made by
empirically oriented schools. Optimality Theory is a response to
functionalism, Principles-and-Parameters a response to typology. What
has remained constant is that generativism still allows only one class
of explanation-- a genetic or biological one. Whatever universal
patterns in language structure are discovered, these are assumed
necessarily to be features of the innate human language ability, and the
ability itself is thought of as a body of knowledge of these strutures.
Functional explanations (economy or efficiency of communication) or
material ones (limitations on vocal/aural apparatus, or on memory) are
ruled out a priori as unworthy or uninteresting.

    In short I am highly sceptical of the way in which biological
analogies have been used in linguistics.  But of course I don't blame
biologists for this. The ultimate problem, as Dr. Ghesiln has hit upon,
is the ontology of language, that is to say, what IS it? In asking this
question to a bunch of linguists, our colleague is soon to discover, if
he hasn't already, that he has stumbled in among, not a group of sober
specialists, but a crowd of blind beggars arguing angrily about the
nature of an elephant.

    Well, here I will foolishly join this congress of the blind by
offering my own minority opinion: Language, as I see it (or as I grope
it), is neither an organ nor an organism, but a cultural artifact or a
tool. Our ability to learn, to use, even possibly to construct from
nothing this tool is due to genetically determined neurological
abilities unique to humans.  But the structural properties of the tool
are independent of the neurological structures which make it possible to
learn or use the tool. People have an innate ability, beyond that of
other animals, to build dwellings and structures of various types, and
these structures tend to have certain common structural properties-- for
example, a roof, something to support the roof. These properties can be
described, and 'explained' to some degree in functional (to keep the
rain out) and physical (gravity) terms.  The ability to build the
structure in the first place is rooted in the brain and is a product of
evolution. The question is, how does the genetic ability to construct an
artifact relate to the general or universal properties of the artifact.
It seems odd to me to assert, as most linguists do, that these things
are simply identical-- to assume in effect that the ability to build
things is encoded in the brain as innate knowledge of a set of universal
structural properties of buildings (the roof principle, the support
principle).  I conclude that analysis of structural properties of the
artifact will not necessarily yield any insight into the neurological
structures which construct or control the artifact, nor will knowledge
of the neurological structures necessarily provide a complete
explanation of the structural properties of the artifact (since
functional and material factors may also be relevant).

    At the risk of introducing yet another false language/biology
analogy, I'll close with a question. Language is possibly more analogous
to anthills, beehives, and birds' nest, than it is to ants, bees, or
birds. How do biologists now conceive the relationship between the
structural properties of beehives and the neurological structure of
bees? Do ants and bees have little blueprints in their ganglia? Is there
variation and change in the structure of beehives and anthills? If so,
can this be described in terms of selectional pressures, and, if so, is
this independent of mutation and speciation of the organisms involved,
or does it necessarily go in tandem with it?

-RR

Ghiselin, Michael wrote:

> ----------------------------Original
> message----------------------------
>                Thanks for your very thoughtful commentary.  What you
>           say points out the fact that when we start asking what some
>           of these fundamental units are, they become increasingly
>           problematic; and when we try to compare across kinds of
>           systems the parallels are evident, but they too are
>           problematic.
>                The problems of what a species is and of what a
>           language is are not unusual.  Indeed I would be surprised to
>
>           find a theoretical term in any science that is not hard to
>           define in a way that pleases all the practitioners.  Small
>           wonder then, that we cannot easily find exact parallels
>           between the fundamental units of interest to linguists and
>           to biologists.  We can say that there are phonemes, words
>           etc., and we can say that there are nucleotide pairs ...,
>           but what we are looking at is hierarchical structure without
>
>           exact functional correspondence.  Geneticists do not agree
>           as to what a gene is, though they work with them and talk
>           about them all the time.
>                One point that I still would like clarified is the
>           relationship between the speaker of the language and the
>           language itself.  The speaker is a part of a language
>           community and the vocabulary, grammar etc. are parts of the
>           language.  The speakers may be said to know, understand,
>           speak, participate in, etc. the language.  But we usually do
>
>           not call them parts of the language.  There are a whole
>           range of related problems with respect to culture in
>           general.  The way Tylor defined "culture" it includes
>           concrete artifacts.  There must be an extensive literature
>           on such issues.  But such material as I have read (including
>
>           the 1952 review by Kroeber and Kluckhohn) does not really
>           face up to the ontological issues.
>           MG

--
+++++++++++++++++++++++++++++++++++++++++++++++++
Robert R. Ratcliffe
Senior Lecturer, Arabic and Linguistics,
Dept. of Linguistics and Information Science
Tokyo University of Foreign Studies
Nishigahara 4-51-21, Kita-ku
Tokyo 114 Japan