Deep relationships among California languages

Abstract

The Hokan and Penutian language classifications, introduced by Dixon & Kroeber (1913), remain controversial nearly a century after they were first proposed. Recently developed computational methods for identifying historical relationships between languages are promising tools for assessing distant linguistic relationship proposals such as these. This paper uses a variation of the linguistic relatedness metric and multilateral clustering procedure developed by Kessler (1999, 2001) to study California language phylogenetics. The purpose is twofold: to evaluate the utility of this methodology for identifying deep relationships, and to re-examine the evidence for Hokan and Penutian groupings. While this paper illustrates several advantages of the methodology employed, it ultimately fails to provide any additional support for Hokan or Penutian. I conclude that while this result may be influenced by the sensitivity of the methodology to the composition of the input sample, it ultimately casts further doubt on the genealogical nature of the Hokan and Penutian classificatory groups.