CIM Seminar with Harald Hammarström

Abstract
A common view holds that the languages in the world, i.e.,
natural languages such as English, Swedish, Norwegian etc, cannot be
enumerated in a scientific manner. Under this view, languages, like,
e.g., religions, may form continua that defying discretization and
counting. In particular, intelligibility between languages is held to
be a gradient property with no non-arbitrary cut-off point. We will
argue that this view is premature. With some simple assumptions on
what consitutes a language, we show that intelligbility can be defined
as a yes/no property in a simple formal way, without imposing an
arbitrary threshold. This formal definition is difficult to apply
in practice, but can be approximated. Further, like in reality,
this binary intelligibility relation is not transitive, so chains
where neighbouring languages are intelligible but the endpoints are not
(``dialect chains'') may form, and pose another challenge to counting.
We show that this latter challenge can be solved with graph-colouring,
yielding a specific number of languages but not necessarily a unique
discretization. Finally, we give an empirical estimate of the number
of languages in the world given the current state of knowledge.

Registration
Please click here to register!