I would like to propose an answer to the general question of how new varieties of a language emerge, and what then happens to them over time. My example is American English, and I have discussed two competing models for the formation of American English (2002: 237), the traditional metaphors of the melting pot and the mixing bowl, concluding that the best solution to explain the historical and social as well as the linguistic facts is that the American colonies represented a melting pot of

more »

... istic features overall, out of which different localities eventually created a new mixing bowl of regional voices. We see that linguistic features were retained from the habits of individual speakers, but not whole linguistic systems from constituent immigrant languages or dialects. The default condition for English in the colonies in the seventeenth century was not "London standard" (whatever that could have meant given the great population mobility of the time), but instead a pool of linguistic features collected from a radically mixed settlement population. What I could not describe at the time was the process by which particular linguistic features came to prevail in the English of the colonies, and by which different features could become more common in different areas. Since then, I have written extensively about the fact that language in use, speech as opposed to linguistic systems as usually described by linguists, satisfies the conditions for complex systems as defined in sciences such as physics, evolutionary biology, and economics (most thoroughly in Kretzschmar 2009). This essay discusses implications for the initial formation of American English and its varieties as the product of random interactions in a complex system between speakers of different input varieties of British English. The findings of this approach contrast sharply with more traditional accounts, but they are not in conflict with Edgar