By Karl C. Diller
Human language provides us a challenging case for the evolution of complexity because this complexity evolves on several levels. Languages are shared cognitive systems. They are hosted in individual human brains, but language systems evolve as people learn the languages and use them to communicate and to express new thoughts. People are not free to introduce innovations in the language systems unless these innovations are immediately understandable through metaphoric or metonymic processes. "Languages," such as English and Dutch, are so variable and changing that Chomsky has argued that they are mere "epiphenomena" that cannot be defined without recourse to "obscure socio-political and normative factors" (Chomsky 1988, 36-37). He asks how we can distinguish between Dutch and German when dialects along the border are so similar. The chief conceptual shift of Chomsky's generative grammar was to focus on language as "an individual phenomenon, a system represented in the mind/brain of a particular individual." I argue elsewhere, however (1993; 1994;1995), that languages are like evolving systems of artificial life, and that using the concepts of complexity and artificial life we can begin to define "languages" with linguistic criteria.
In this paper I will discuss the evolution of linguistic complexity on three levels: the evolution of individual language systems from one-word utterances to complex usage as children learn their native languages and as adults learn second languages; the evolution of simple pidgin languages to the more complex creole languages, through which we see some of the biological constraints on languages (what Bickerton (1984) calls the "bioprogram"); and the evolution of mature languages in the process of "grammaticalization," in which lexical items (such as the verb go) come to be used as grammatical markers (as in the future marker going to) (Hopper and Traugott 1993). We will see that languages are not static systems either in the behaviorist sense of descriptive structural linguistics where there were conditioned habits, or in the Chomskian generative sense where there are grammatical algorithms within idealized individuals. Instead, languages are non-linear, dynamic, and increasingly complex both in the individual and in the (partially) shared systems of language communities.