Linguists Tested 191 Universal Grammar Rules. Only One-Third Survived

A new study finds that one-third of proposed grammatical “universals” hold up under rigorous testing.

Although the world’s languages differ enormously in sound systems, vocabulary, and structure, researchers have long observed that certain grammatical patterns appear repeatedly across cultures. A new study finds that many of these recurring features may be more than a coincidence.

After applying advanced evolutionary modeling techniques, the researchers report that about one-third of the long-proposed “linguistic universals”—patterns believed to exist across all languages—show clear statistical support.

The international research team, led by Annemarie Verkerk (Saarland University) and Russell D. Gray (Max Planck Institute for Evolutionary Anthropology), analyzed data from Grambank, the most extensive global database of grammatical features. They tested 191 proposed universals across more than 1,700 languages, making this one of the broadest examinations of cross-linguistic structure to date.

Moving Beyond Simple Sampling

In earlier research, linguists often tried to avoid bias by selecting languages from distant regions or unrelated families in order to minimize shared ancestry or regional influence.

However, simply sampling widely separated languages does not eliminate hidden historical or geographic connections, and it can weaken statistical reliability while offering little insight into how grammatical traits evolved.

To address these limitations, the team used Bayesian spatio-phylogenetic analyses that explicitly account for both genealogical relationships and geographic proximity among languages. This approach provides a level of statistical rigor rarely achieved in previous work, offering stronger evidence that some grammatical patterns truly recur across human languages rather than arising from shared history alone.

Languages do not evolve at random

“In the face of huge linguistic diversity, it is intriguing to find that languages don’t evolve at random,” says Verkerk. “I am delighted that the different types of analyses we did converged on very similar results, suggesting that language change must be a central component in explaining universals.”

The study found strong evidence for patterns involving word order (such as whether verbs precede or follow objects) and hierarchical universals (such as dependencies in which arguments are marked in grammatical agreement). The patterns predicted by the supported universals have evolved repeatedly across the world’s languages, suggesting deep-rooted constraints in how humans structure communication.

Senior author Russell Gray reflected, “We discussed whether to write this up as a glass-half-empty paper — ‘look how many proposed universals don’t hold’ — or a glass-half-full paper — ‘there’s robust statistical support for about a third.’ In the end, we chose to highlight the patterns that evolve repeatedly, showing that shared cognitive and communicative pressures push languages toward a limited set of preferred grammatical solutions.”

By demonstrating which universals truly hold up under evolutionary scrutiny, the study narrows the field for future research into the cognitive and communicative foundations of human language.

Reference: “Enduring constraints on grammar revealed by Bayesian spatiophylogenetic analyses” by Annemarie Verkerk, Olena Shcherbakova, Hannah J. Haynie, Hedvig Skirgård, Christoph Rzymski, Quentin D. Atkinson, Simon J. Greenhill and Russell D. Gray, 17 November 2025, Nature Human Behaviour.
DOI: 10.1038/s41562-025-02325-z

Never miss a breakthrough: Join the SciTechDaily newsletter.
Follow us on Google and Google News.