Document Type



Language—as with most communication systems—likely evolved by means of natural selection. Accounts for the genetical selection of language can usually be divided into two scenarios, either of which used in isolation of the other appear insufficient to explain the phenomena: (1) there are group benefits from communicating, and (2) there are individual benefits from being a better communicator. In contrast, it is hypothesized that language phenotypes emerged during a coevolutionary struggle between parental genomes via genomic imprinting, which is differential gene expression depending on parental origin of the genetic element. It is hypothesized that relatedness asymmetries differentially selected for patrigene-caused language phenotypes to extract resources from mother (early in development) and matrigene-caused language phenotypes to influence degree of cooperativeness among asymmetric kin (later in development). This paper reports that imprinted genes have a high frequency of involvement in language phenotypes ( 36%), considering their presumed rarity in the human genome ( 2%). For example, two well-studied genes associated with language impairments (FOXP2 and UBE3A) exhibit parentof- origin effects. Specifically, FOXP2 is putatively paternally expressed, whereas UBE3A is a maternally expressed imprinted gene. It is also hypothesized that the more unique and cooperative aspects of human language emerged to the benefit of matrilineal inclusive fitness. Consistent with this perspective, it is reported here that the X-chromosome has higher involvement in loci that have associations with language than would be expected by chance. It is also reported, for the first time, that human and chimpanzee maternally expressed overlapping imprinted genes exhibit greater evolutionary divergence (in terms of the degree of overlapping transcripts) than paternally expressed overlapping imprinted genes. Finally, an analysis of global language patterns reveals that paternally but not maternally silenced Alu elements are positively correlated with language diversity. Furthermore, there is a much higher than expected frequency of Alu elements inserted into the protein-coding machinery of imprinted and X-chromosomal language loci compared with nonimprinted language loci. Taken together these findings provide some support for parental antagonism theory. Unlike previous theories for language evolution, parental antagonism theory generates testable predictions at the proximate (e.g., neurocognitive areas important for social transmission and language capacities), ontogenetic (e.g., the function of language at different points of development), ultimate (e.g., inclusive fitness), and phylogenetic levels (e.g., the spread of maternally derived brain components in mammals, particularly in the hominin lineage), thus making human capacities for culture more tractable than previously thought.