The ideas of the origins of language group have first been tested in
a number of simulations. Although the ultimate goal is to ground the experiments
in the physical world by implementing them on robots
that operate in controlled, but real environments, the development of the
ideas can be done much quicker in simulation.
The simulations are based on the methodology of artificial life. They consist
of populations of agents that play adaptive "language" games
with each other. Simulations have been performed for the development of
phonology, lexicon,
(spatial) meaning and for the co-evolution
of words and meanings. At the moment, simulations for the development
of syntax are under construction.
Simulations are typically implemented in LISP on macintoshes or
in C++ on PC's. Population size is usually somewhere between 10-50 agents
and the games are iterated between 1000-100 000 times.
Back to Top
