"Ancient DNA shared with Neanderthals may explain human language (ScienceDaily, 12 June 2026). Researchers from University of Iowa Health Care examined a class of non-coding DNA elements called Human Accelerated Quantitative Enhancer Regions (HAQERs), regulatory sequences that influence gene expression in the brain.
By correlating variation in these enhancers with differences in present-day language ability, the team argues that some of the genetic architecture underlying language predates the divergence of modern humans and Neanderthals.
“What we’re seeing is how a very small part of the genome can have an outsized influence, not just on who we were as a species, but on who we are as individuals,” Michaelson [one of the authors of the journal article] says, noting that HAQERs represent less than a tenth of a percent of the genome but drive roughly 200 times more impact on language ability than any other genomic region.
The study does not identify a “gene for language,” nor does it uncover a new biological mechanism.
*HAQERs are a subclass of already known regulatory DNA,
.webp)
*These DNA elements do not code for proteins; they function by modulating the activity of other genes, often at a distance. Over the past two decades, such regions have moved from the margins of genomics to the center, as the old notion of “junk DNA” has given way to the far more intricate picture of genome-wide regulation.
That picture has only grown more complex and integrated with the rise of the “4D genome.” DNA is not a linear string but a folded, dynamic structure. Enhancers like HAQERs exert their effects through spatial proximity, looping across the genome to contact target genes, and their activity depends on timing, cell type, and developmental context. In such a system, causation is distributed and non-local.
From 4D Reality to 1D Statistics
What is new in this study is not the existence of these elements, but the statistical association drawn between their variation and language performance in living populations. From this, the authors infer deep evolutionary continuity, projecting present-day correlations back into an ancestral narrative spanning millions of years.
This brings us to the journal paper itself:
Ancient regulatory evolution shapes individual language abilities in present-day humans, (Science Advances, Lucas G. Casten, et al., 22 April 2026).
The authors draw this deep time inference with a brand new statistical method, pioneered specifically by these scientists for this paper. It is called an Evolutionary Stratified Polygenic Score (ES-PGS):
"To investigate the genetic origins of language ability, we developed an ES-PGS approach that systematically examines how genetic variants from different evolutionary periods contribute to traits."
In essence, ES-PGS is a statistical tool that aggregates many small genetic correlations with a trait, in this case, language-related performance, and then groups those correlations according to when the underlying DNA variants are thought to have arisen/emerged in evolutionary history. By comparing genomes across species and fitting them to a phylogenetic model, the method assigns variants to different “time layers,” some stretching back tens of millions of years.
This procedure does not directly observe evolutionary events.
It begins with present-day genetic variation and, using a model of evolutionary relationships and assumed timelines, it infers when particular variants likely emerged.
In essence, we are presented again with the ever-present circularity and self-reinforcement of evolutionary reasoning. ES-PGS partitions present-day genetic correlations according to an assumed evolutionary timeline, and then reports the results as evidence of that timeline."
CEH