Neutral quasispecies evolution and the maximal entropy random walk

Even if they have no impact on phenotype, neutral mutations are not equivalent in the eyes of evolution: A robust neutral variant—one which remains functional after further mutations—is more likely to spread in a large, diverse population than a fragile one. Quasispecies theory shows that the equilibrium frequency of a genotype is proportional to its eigenvector centrality in the neutral network. This paper explores the link between the selection for mutational robustness and the navigability of neutral networks. I show that sequences of neutral mutations follow a “maximal entropy random walk,” a canonical Markov chain on graphs with nonlocal, nondiffusive dynamics. I revisit M. Smith’s word-game model of evolution in this light, finding that the likelihood of certain sequences of substitutions can decrease with the population size. These counterintuitive results underscore the fertility of the interface between evolutionary dynamics, information theory, and physics.

Factors driving lexical variation in L2 French: A variationist study of automobile, auto, voiture, char and machine

ABSTRACTOur paper examines lexical variation in the spoken French of second language learners and focuses on words referring to the notion of ‘automobile’ (i.e., automobile, auto, voiture, char and machine). Results reveal that while students do follow the native speaker pattern of using the neutral variant auto in most instances, they diverge from native speakers by making no use of the vernacular form char and relatively high use of the prestige variant voiture. The principal external factors that influence variant choice are students' home language and the representation of variants in the input to which students are exposed.