Matching the simulations to the situation: A synthesis of computational modelling and archaeological data.

Ben Marwick

Centre for Archaeology, University of Western Australia

benm@cyllene.uwa.edu.au


Two of the most productive, and until now, independent areas of inquiry
in human language evolution are computational linguistics and
archaeology.  Computational linguists have produced a series of
simulations modelling non-biological evolutionary processes.  These
simulations reveal important details in the probable stages of language
evolution and the significance of variables such as population size and
learning period.  Archaeological data provides opportunities to test the
computational models by using the models as interpretive frameworks for
archaeological evidence.  In this paper I show that distances of
raw-material transportation throughout the Plio-Pleistocene reflect how
hominid groups gather and exchange information.  Early hominids moved
raw materials short distances, suggesting a home range size, social
complexity and communication system similar to primates in equivalent
environments.  After about 1.0 million years ago there was a large
increase in raw-material transfer distances, possibly a result of the
emergence of the ability to pool information by using a protolanguage. 
Another increase in raw-material transfer occurred during the late
Middle Stone Age in Africa (after about 130,000 years ago), suggesting
the operation of exchange networks.  Exchange networks require a
communication system with syntax, the use of symbols in social contexts
and the ability to express displacement, which are the features of human
language.  These developments can be interpreted using the results of
simulations of non-biological evolution of syntax.  In this paper I show
how this archaeological data can be interpreted using the results of
computational models to produce a robust and testable narrative of human
language evolution.