Models of Language Evolution: Does the Math Add Up?
Bart de Boer, Willem Zuidema

Abstract:
The last two decades have seen an explosion of interest in mathematical
and computational models of language evolution. Formal modelling is
seen by increasingly many in the field as an approach that ensures
internal consistency of evolutionary scenarios.  However, there has
been little attention to the question of how well the many different
models fit together. Are they consistent with and complementary to each
other? Is there a series of models that really covers the evolutionary
emergence of modern language from a prelinguistic, ancestral state? Are
the assumptions that go into a particular model, if not (yet) supported
by empirical findings, made plausible by results from other models?

In this paper, we argue that these problems deserve much more attention
than they currently receive. For sustaining the success of modelling
approaches in language evolution research, it is crucial that models
start living up to their promise: modellers must make explicit how their
models fit in with other work in complete scenarios on the origins
of language(s), and how their modelling results affect judgments of
plausibility of one scenario against another. Moreover, they must do so
based on careful consideration of other work, without overstating their
results and misusing the prestige that comes with mathematical approaches.

Our arguments are based on a particular view on the role of modelling in
scientific research in general, and in "historical" research fields with a
paucity of direct evidence in particular. In section 2, we will therefore
start with some considerations about the methodology of modelling in
our field. To ground the discussion, however, we will quickly move to
concrete examples. In section 3 we will discuss the contributions and
shortcomings of models of the evolution of speech. In section 4 we will
then draw some general lessons from this case study, and sketch an agenda
for future research in the language evolution modelling field at large.