Current work on the relationship between languages tends to be in terms either of a family tree model of ‘parent-and-child’ linkage, or of a linguistic area model of the diffusion of categories and forms between adjacent languages. Family tree and linguistic area are most often treated as distinct phenomena; in this essay I will attempt to integrate them within a global view of linguistic development.

A language does not exist in a vacuum but is the means for communication within a group of people, with a certain political and economic system. An integrated theory of language development must pay attention to the way of life of the group speaking a known language, or the postulated earlier stage of a language, and their political, social and linguistic relationships with neighbouring groups speaking other languages.

The family tree model was developed for – and is eminently appropriate to – the Indo-European (IE) language family. It has become the received view of how languages are related, so that linguists attempt to discover an IE-like family tree structure in every group of languages, from anywhere in the world (whatever their typological profile); along with this goes an attempt at detailed subgrouping. But in many cases there is no serious attempt to reconstruct part of the system of a proto-language (which is needed, as proof of a family-tree-type genetic relationship). For some groups of languages – for instance, Semitic and Polynesian – the family tree model is entirely applicable. For others it may be less so; the similarities that have been taken as evidence for genetic relationship may really be due to areal diffusion.