The Neolithic Hypothesis

[Ray Hendon]

The discussion you guys are having now reminds me of Earnest Hemmingway's
wonderfully simple phrase from Islands in the Sun, when he had the dying
hero say to his faithful companion, "They are all right."  He meant, I
believe, that on any issue there will be contention and that all sides are
correct--there is truth on all sides.

    A somewhat contentious issue in the community of linguists, at least for
what I can see from my email, is that of making predictions.  I read much
that says "It is not the job of linguists to provide prediction as to what
will happen, but only to describe what we know has happened."  The
contention of this group is that prediction is entirely outside the ken of
linguists.  Description is king; prediction is for other fields. These folks
are anti-prognosticators or anti-advocates of prediction.

    A second group, less unforgiving in its disdain of quantitative methods,
are the non-advocate group, the agnostics.  The agnostics do not advocate
prognostication, but their objections are not based on the belief that
predicting is bad, just that it is wrong.  We do not know enough, they say,
to predict what will happen in any language over time, and thus attempts at
predicting are bound to fail.  Lack of sufficient information is the basis
of their concern.  Their focus in on analyzing existing languages, but not
on predicting what will happen in the future.

    There is certainly truth in the contention of the non-advocate
agnostics.  The simple truth is we cannot know how things will change in the
future that are both exogenous to our culture and beyond our control .  As
was pointed out, we all speak a slightly different language than our fathers
and mothers spoke when they were our age.  When I tell my 88 year-old father
about my email conversations, he has some idea of what I mean, but I doubt
he has ever used the word "email" in his life.  There is no way he could
have predicted even fifty years ago that the word "email" would one day
enter his vocabulary.  My suggestion for the applicability of quantitative
methods is not applicable for all domains of linguistics.  I leave this
category of speculation and projection to those with more insight than I.
Metaphysics has never been my strength, and I generally agree with those who
assert that some things are too complex to predict.

     A third group of linguists could be called pro-advocacy-lite.  Their
advocacy of prediction is constrained to the macro domain and largely to the
past.  They accept that predicting linguistic development may be done, but
it can be done only at an aggregated level and of past events. The followers
of the IE hypothesis, for example, say that their model of linguistic
development is sufficiently informed so as to predict what language you
would hear spoken if you were to go back to 7000 BCE in a certain area in
Europe/Asia and listen. They can predict how a certain word would be
pronounced  or used in Sumaria in 3000 BCE,  or Rome in 250 BCE, or London
in 1070 ACE.  They believe in the inherent possibility that language can be
predicted, or at least certain aspects can be predicted.  But they are
wisely skeptical of moving their predictions into the future and prefer to
keep the time-line of their prediction facing the past.

      But there may be another group of linguists who work in a domain that
overlaps some but not entirely with other groups.  This domain deals with
the adoption of a language within a specific geographical area at a certain
time.  A macro-orientation, dealing with an aggregated concept of
language--collections of words that constitutes Old Saxon, Middle English,
etc.   This domain takes as its ken the interfacing of languages.  The
movement of a language into an area, the interaction and intersection of two
or more languages within a given population are their subjects.  This domain
is not concerned so much with changes within a language, but the adoption of
a specific language by a specific population.  History is full of episodes
of invaders taking their language with them and imposing it upon the conquer
people.  And, history is full of peoples who absorb a linguistic invasion
without changing the fundamental structure of their mother tongue.  It is
also filled with examples where a new language develops from parts of many
other languages.  In the adoption domain I propose that a quantitative model
or prediction may hold some promise as a predictive tool.

To this last group I still urge skepticism about prediction, but skepticism
of both sides.  Be skeptical of the proposition, "predictions can be made,"
but also be skeptical of the proposition that "all predictions are wrong."
There are some things that we can predict with great precision.  Perhaps
linguistics is sufficiently mature now, thanks to the hundreds of years of
work done by earlier scholars, to begin using tools of other sciences and
take up the ultimate challenge of all scientific study: predict as well as
explain.  As long as it know the limits of how its predictions will fare,
and into which domain to confine its inquiry,  I do not see in the existing
arguments reasons for not trying. To the mathematician and statistician, it
doesn't matter which direction you face.  You can build a model that deals
with expectations of both the future and the past.  In one sense, it doesn't
matter which because the same procedures are used for both directions.

Regards,
Ray Hendon