As palavras funcionando como alelos: conectando a evolução da linguagem com aprendizes bayesianos de modelos de deriva genética

quarta-feira, dezembro 23, 2009

Words as alleles: connecting language evolution with Bayesian learners to models of genetic drift

Florencia Reali* and Thomas L. Griffiths
- Author Affiliations

Department of Psychology, 3210 Tolman Hall, MC 1650, University of California at Berkeley, Berkeley, CA 94720-1650, USA

* Author for correspondence (


Scientists studying how languages change over time often make an analogy between biological and cultural evolution, with words or grammars behaving like traits subject to natural selection. Recent work has exploited this analogy by using models of biological evolution to explain the properties of languages and other cultural artefacts. However, the mechanisms of biological and cultural evolution are very different: biological traits are passed between generations by genes, while languages and concepts are transmitted through learning. Here we show that these different mechanisms can have the same results, demonstrating that the transmission of frequency distributions over variants of linguistic forms by Bayesian learners is equivalent to the Wright–Fisher model of genetic drift. This simple learning mechanism thus provides a justification for the use of models of genetic drift in studying language evolution. In addition to providing an explicit connection between biological and cultural evolution, this allows us to define a ‘neutral’ model that indicates how languages can change in the absence of selection at the level of linguistic variants. We demonstrate that this neutral model can account for three phenomena: the s-shaped curve of language change, the distribution of word frequencies, and the relationship between word frequencies and extinction rates.

language evolution genetic drift Bayesian inference neutral models


↵1 Our description corresponds to a ‘haploid’ version of the Wright–Fisher model, with just one allele per organism. In the more conventional diploid model, an additional factor of 2 appears in front of N, as N is taken to be the number of organisms rather than the number of alleles.

↵2 Note that, under certain conditions, the shape of the stationary distribution does not correspond exactly to the shape of the prior. For example, as illustrated in figure 1 for the case of K = 2 and N = 10, a uniform prior given by α/2 = 1 is associated with a bell-shaped stationary distribution.

↵3 As suggested by an anonymous reviewer, it is possible that the model in Baxter et al. (2006) admits an extension to accommodate the infinite case, but this possibility has not been explored in previous work.

↵4 In our example we assume that languages can only maintain one kind of word order (and that there are only two possibilities). This is simplifying assumption, as illustrated by languages such as German where word order in subordinate clauses can differ from word order in main clauses.

Received August 21, 2009.

Accepted September 21, 2009.
© 2009 The Royal Society