Your Voice Box Shapes How Languages Diverge, Study Finds

Every human voice is unique, and that uniqueness may run far deeper than the sound of an individual’s name being called across a room. A new study published in PLOS Biology suggests that the physical differences in our vocal tracts—differences as ordinary as the shape of our palates, the length of our pharynxes, or the size of our tongues—leave measurable fingerprints on the way languages sound. Over generations, those fingerprints may help drive the astonishing phonetic diversity found among the world’s roughly 7,000 spoken languages.

The research, conducted by Steven Moran, Tiena Danner, Marcia S. Ponce de León, and Christoph P. E. Zollikofer, builds on a deceptively simple observation: human vocal tracts, like other physical traits such as height or skull shape, vary considerably from person to person. What has remained unclear is how this anatomical variation translates into phonetic variation between speakers, and whether that individual-level variation can accumulate into the large-scale differences we hear between languages and language families.

To answer that question, the team constructed what they call a biocultural model of phonetic evolution. The model is explicitly modeled on a cornerstone framework in evolutionary biology: Russell Lande’s model of multivariate phenotypic evolution, which describes how multiple quantitative traits within a population change over time under the combined forces of mutation, selection, and genetic drift. In the researchers’ analogy, individual vocal tract morphologies play the role of genetic variation within a population, while the acoustic and phonetic output of those anatomies plays the role of the expressed phenotype.

The core finding is that individual vocal tract shapes, together with the articulatory strategies speakers adopt to produce speech, leave consistent acoustic signatures. Two speakers may intend to produce the same vowel—the same phonological target—but their distinct anatomies cause the acoustic result to differ in patterned, predictable ways. Crucially, the characteristic pattern of phonetic variation found within groups of speakers correlates with the pattern of phonetic variation observed between groups. In other words, the raw biological variability inside a community mirrors, in its structure, the variability that distinguishes communities from one another.

This correlation invokes one of the central tenets of evolutionary biology: within-group variation is the raw material for between-group diversification. Just as genetic drift acting on standing variation within populations can push separated groups toward different trait values, the researchers propose that phonologically similar vowel systems can drift toward phonetically distinct realizations through a neutral-like evolutionary process. Two languages might share essentially the same vowel inventory on paper—five vowels, say, arranged in the same phonological categories—yet over many generations of speakers, the actual acoustic realizations of those vowels can diverge simply because different speaker populations carry different distributions of vocal tract morphologies.

The technical heart of the study lies in how the model links anatomy to acoustics. The geometry of the vocal tract—the tube stretching from the vocal folds to the lips—determines its resonance properties, which in turn shape the acoustic output of any articulatory gesture. Because these resonances, known as formants, depend continuously on tract dimensions, even modest anatomical differences produce measurable shifts in vowel quality. By simulating how populations of speakers with varying tract morphologies produce and transmit phonetic targets across generations of learners, the researchers could track whether the resulting distribution of sounds within a group predicts the distribution between groups, and the answer was yes: the covariance structure is preserved across levels.

Perhaps the most consequential use of the model is as a null hypothesis. In evolutionary studies, a neutral model describes what happens in the absence of adaptive forces, allowing researchers to detect when selection has been at work. Here, the neutral-like process would be the passive accumulation of anatomically driven phonetic variation. If real languages show phonetic patterns that deviate from this neutral expectation, the deviation may signal culturally mediated language change—processes such as norms of prestige, intentional differentiation between communities, or contact-driven convergence and divergence. The model thus gives linguists and speech scientists a principled baseline for disentangling biological mechanisms from cultural ones, a distinction that has long been difficult to draw with rigor.

The implications stretch across several disciplines. For biolinguistics, the study suggests that the biology of the speaker is not merely a neutral channel through which language passes, but an active source of the variation on which language change operates. For phonetics and speech science, it offers a framework for predicting how anatomical diversity within a population shapes the statistical landscape of its speech sounds. And for historical linguistics, it raises the possibility that some portion of the phonetic diversity among related languages, often attributed to cultural or contact histories, could instead reflect neutral drift seeded by differences in speaker anatomy across populations.

It is important to note what the model does not claim. The finding does not suggest that anatomy determines language, or that speakers of different ancestries are constrained to produce different phonological systems. The world’s languages demonstrably cross-cut any anatomical grouping of their speakers, and any healthy human can learn any human language. What the model proposes is subtler: given identical phonological targets, different distributions of vocal tract morphology across speaker groups bias the acoustic outcomes in a patterned way, and over long timescales, that patterning can compound into detectable phonetic divergence even without any cultural impetus.

The researchers frame their biocultural model as a starting point rather than an endpoint. By providing an explicit, quantitative null expectation for how phonetic diversity should evolve from human biological variation alone, the model enables future work to test specific languages and sound changes against that baseline, flagging the cases where culture demonstrably bends the trajectory of speech. If the approach holds up across larger datasets and more sound classes, it may quietly resolve one of the oldest debates in the language sciences: how much of the way we speak belongs to our bodies, and how much belongs to our histories. The new answer is that the two are intertwined, with our voices supplying the variation from which our languages are made.

Subject of Research: How interindividual vocal tract anatomical variation drives phonetic diversification across spoken languages

Article Title: Interindividual vocal tract diversity influences the phonetic diversification of spoken languages

Article References: Moran, S., Danner, T., Ponce de León, M. S., & Zollikofer, C. P. E. (2026). Interindividual vocal tract diversity influences the phonetic diversification of spoken languages. PLOS Biology, 24(10), e3004003. https://doi.org/10.1371/journal.pbio.3004003

Image Credits: AI Generated

DOI: 10.1371/journal.pbio.3004003

Keywords: vocal tract, phonetics, phonetic diversity, biocultural evolution, vowel systems, PLOS Biology, speech acoustics, language change, multivariate evolution, anatomy, linguistics, null hypothesis