From the Cover: INAUGURAL ARTICLE by a Recently Elected Academy Member:A comparison of worldwide phonemic and genetic variation in human populations

Worldwide patterns of genetic variation are driven by human demographic history. Here, we test whether this demographic history has left similar signatures on phonemes—sound units that distinguish meaning between words in languages—to those it has left on genes. We analyze, jointly and in parallel, phoneme inventories from 2,082 worldwide languages and microsatellite polymorphisms from 246 worldwide populations. On a global scale, both genetic distance and phonemic distance between populations are significantly correlated with geographic distance. Geographically close language pairs share significantly more phonemes than distant language pairs, whether or not the languages are closely related. The regional geographic axes of greatest phonemic differentiation correspond to axes of genetic differentiation, suggesting that there is a relationship between human dispersal and linguistic variation. However, the geographic distribution of phoneme inventory sizes does not follow the predictions of a serial founder effect during human expansion out of Africa. Furthermore, although geographically isolated populations lose genetic diversity via genetic drift, phonemes are not subject to drift in the same way: within a given geographic radius, languages that are relatively isolated exhibit more variance in number of phonemes than languages with many neighbors. This finding suggests that relatively isolated languages are more susceptible to phonemic change than languages with many neighbors. Within a language family, phoneme evolution along genetic, geographic, or cognate-based linguistic trees predicts similar ancestral phoneme states to those predicted from ancient sources. More genetic sampling could further elucidate the relative roles of vertical and horizontal transmission in phoneme evolution.Both languages and genes experience descent with modification, and both are affected by evolutionary processes such as migration, population divergence, and drift. Thus, although languages and genes are transmitted differently, combining linguistic and genetic analyses is a natural approach to studying human evolution (1, 2). Cavalli-Sforza et al. (3) juxtaposed a genetic phylogeny with linguistic phyla proposed by Greenberg (described in ref. 4) and observed qualitative concordance; however, their comparison of linguistic and genetic variation was not quantitative. A later analysis of genetic polymorphisms and language boundaries suggested a causal role for language in restricting gene flow in Europe (5). More recently, population-level genetic data have been compared with patterns expected from language family classifications (2, 6–12). Other studies addressed whether the serial founder effect model from genetics—human expansion from an origin in Africa, followed by serial contractions in effective population size during the peopling of the world (13, 14)—explains various linguistic patterns (15–19).Past studies are generally asymmetrical in their approaches to the comparison of genes and languages: some focus on genetic analysis and use linguistics to interpret results, and others analyze linguistic data in light of genetic models. Our study directly compares the signatures of human demographic history in microsatellite polymorphisms from 246 worldwide populations (20) and complete sets of phonemes (phoneme inventories) for 2,082 languages; these are the largest available datasets of both genotyped populations and phonemes, the smallest units of sound that can distinguish meaning between words. Languages do not hold information about deep ancestry as genes do, and phoneme evolution is complex: phonemes can be transmitted vertically from parents to offspring or horizontally between speakers of different languages, and phonemes can change over time within a language (21–23). We compare the geographic and historical patterns evident in phonemes and genes to determine the traces of human history in each data type.Phonemic data were compiled by M.R. (the Ruhlen database); for 2,082 languages with complete phoneme inventories and referenced sources in this database, we annotated each language with geographic coordinates (Fig. 1A) and the number of speakers reported (24). We also analyzed PHOIBLE (PHOnetics Information Base and Lexicon) (25), a linguistic database with phoneme inventories for 968 languages. For 139 globally distributed populations in the Ruhlen database (114 in PHOIBLE), we matched each population’s genetic data to the phoneme inventory of its native language (20), producing novel “phoneme–genome datasets” that allow joint analysis of genes and languages.Open in a separate windowFig. 1.Procrustes-transformed PCs for all phonemes and regional axes of phonemic and genetic differentiation. (A) Locations of 2,082 languages in the Ruhlen database. Phoneme inventory size of each language is indicated by the color bar. We performed Procrustes analyses to compare the first two PCs of phonemic data (B and C) and genetic data (D) to the geographic locations of languages/populations (P < 10⁻⁵ for all three comparisons after 100,000 permutations). The mean Procrustes-transformed PC values (B) for phonemes in the Ruhlen database (t₀ = 0.57), (C) for phonemes in PHOIBLE (t₀ = 0.52), and (D) for allele frequencies (t₀ = 0.69) are displayed in each geographic region. Circle size corresponds to number of languages (B and C) or populations (D). (E) For the Ruhlen phoneme–genome dataset, pairwise geographic distance matrices were projected along different axes (calculated at 1° intervals); within each region, the rotated axis of geographic distance that was most strongly associated (greatest Mantel r) with phonemic distance (black arrows) and genetic distance (gray dashed arrows) is shown. Thinner arrows (Europe, East Asia, South America) indicate nonsignificant associations. Black dots indicate population locations for the Ruhlen phoneme–genome dataset. With the exception of North America, axes of phonemic differentiation and genetic differentiation are similar in most regions (North America: 78° difference; other regions: mean difference 16°).To compare the signatures of human demographic history on genetic variation and phoneme inventories, we used Procrustes analyses to compare principal components (PCs) for both data types with sample geographic locations and determined whether phonemic and genetic distance are more correlated than expected from geographic distance alone. We also developed a new method for identifying regional axes of linguistic and genetic differentiation and tested whether the origin of the human expansion out of Africa can be detected from the geographic distribution of the numbers of phonemes in languages (phoneme inventory sizes). Conflicting predictions exist for the effects of geographic isolation and population contact on language evolution (e.g., refs. 26–29); we tested these by comparing phoneme inventories according to language density at varying radii. We also quantified the extent to which phoneme evolution can be modeled along genetic, geographic, and cognate-based phylogenies. With these joint analyses, we tested whether phonemes and alleles carry signatures of ancient population divergence and recent human migrations, and we identified demographic processes that have different effects on phonemes and alleles.