Triangulation supports agricultural spread of the Transeurasian languages

The paper investigates the origins, timing, and dispersal mechanisms of the Transeurasian (Altaic) languages—Japonic, Koreanic, Tungusic, Mongolic, and Turkic—within the broader context of human, linguistic, and cultural expansions in Northeast Asia. It addresses the long-standing debate over whether similarities among these language families reflect common inheritance or borrowing and challenges the traditional ‘pastoralist hypothesis’ that links their primary dispersals to nomadic steppe expansions in the late fourth millennium BP. Instead, the authors propose a ‘farming hypothesis’ that associates the primary splits and early spread of Transeurasian languages with Neolithic millet-farming populations. Given that the issues extend beyond linguistics alone, the study adopts an explicitly interdisciplinary ‘triangulation’ of linguistics, archaeology, and genetics to test this hypothesis in a region—Northeast Asia—that has been comparatively underrepresented in recent ancient DNA and interdisciplinary research.

The relatedness of Transeurasian (Altaic) languages has long been debated, with many shared features attributed to borrowing, though recent assessments identify a core of reliable evidence supporting a genealogical relationship. Competing homelands have been proposed across Inner Asia and Northeast Asia, including the Altai, Yellow River, Greater Khingan, and Amur regions. Interdisciplinary studies for Northeast Asia remain scarce compared to western Eurasia, with prior work often focusing on single disciplines (notably genetics) or reviews of existing datasets. Recent advances in ancient DNA have reshaped understanding of Eurasian prehistory but leave eastern Eurasia, especially Northeast Asia, less resolved. The paper situates itself against earlier steppe-centered models of language spread and builds upon archaeobotanical research indicating multiple centers of millet domestication (e.g., West Liao and Yellow River basins) and archaeological syntheses of Neolithic and Bronze Age dispersals in the region.

The study triangulates three disciplines with newly compiled datasets and quantitative analyses:

• Linguistics: Compiled a dataset of 3,193 cognate sets covering 254 basic vocabulary concepts across 98 Transeurasian language varieties (including dialects and historical stages). Applied Bayesian phylogenetic methods to infer a dated language tree and Bayesian phylogeography to reconstruct spatiotemporal expansions and homelands; complemented by qualitative analysis of reconstructed agropastoral vocabulary to identify culturally diagnostic items for ancestral communities.
• Archaeology: Extracted 172 archaeological features for 255 Neolithic–Bronze Age sites across northern China, the Amur and Primorye, Korea, and Japan. Compiled 269 direct radiocarbon-dated early crop remains. Performed Bayesian clustering of sites based on cultural similarity to identify spatiotemporal patterns associated with farming dispersals (millet and later rice/wheat), and synthesized demographic trends linked to the spread of agriculture.
• Genetics: Generated genome-wide data for 19 authenticated ancient individuals from the Amur, Korea, Kyushu, and the Ryukyus and integrated them with published genomes (9500–300 BP) spanning the eastern steppe, West Liao, Amur and Yellow River regions, Liaodong, Shandong, Primorye, and Japan. Projected ancient individuals onto principal component analyses of present-day Eurasian and East Asian populations; modeled admixture using qpAdm with components representing Amur-like (Jalainur), Yellow River (Yangshao), Jomon (Rokutsu), and mixed West Liao (Hongshan, Upper Xiajiadian) ancestries. Interpreted ancestry profiles through time to test correspondence with inferred linguistic and archaeological dispersals.

• Time depth and phylogeny: Bayesian inference estimates the Proto-Transeurasian root at 9181 BP (95% HPD 5595–12793); Proto-Altaic (Turkic–Mongolic–Tungusic) at 6811 BP (4404–10166); Proto-Mongolo-Tungusic at 4491 BP (2599–6373); and Proto-Japano-Koreanic at 5458 BP (3335–8024). These represent the initial break-ups into major subgroups.
• Homeland and dispersals: Bayesian phylogeography supports an Early Neolithic origin in the West Liao River region. Following a primary Neolithic split, dispersals proceeded in the Late Neolithic and Bronze Age: Proto-Mongolic northwards to the Mongolian Plateau; Proto-Turkic westwards across the eastern steppe; Proto-Tungusic to the Amur–Ussuri–Khanka region; Proto-Koreanic to the Korean Peninsula; and Proto-Japonic over Korea to the Japanese Islands.
• Reconstructed lexicon: Neolithic-stage proto-languages (Proto-Transeurasian, Proto-Altaic, Proto-Mongolo-Tungusic, Proto-Japano-Koreanic) preserve a core vocabulary linked to cultivation (e.g., ‘field’, ‘sow’, ‘plant’, ‘spade’), millet (e.g., ‘millet seed’, ‘millet gruel’, ‘barnyard millet’), food processing (‘ferment’, ‘grind’, ‘brew’), wild foods (‘walnut’, ‘acorn’, ‘chestnut’), textile production (‘sew’, ‘weave’, ‘loom’, ‘spin’, ‘ramie’, ‘hemp’), and domesticated pigs and dogs—without rice, wheat, barley, or dairying. Bronze Age subfamilies show new terms for rice, wheat, barley, dairying, cattle/sheep/horses, tools, and silk, largely attributable to borrowing during intensified interaction.
• Archaeology: Bayesian clustering of 255 sites reveals a Neolithic West Liao cluster splitting into branches associated with millet farming: to Korea (Chulmun) and to the Amur–Primorye–Liaodong. Millet spread to Korea by ~5500 BP and to Primorye via the Amur by ~5000 BP. Bronze Age sites in West Liao cluster with Mumun (Korea) and Yayoi (Japan), reflecting addition of rice and wheat in the fourth millennium BP, transmission to Korea in the Early Bronze Age (3300–2800 BP), and to Japan after ~3000 BP. Evidence for animal domestication before the Bronze Age is limited (mainly dogs and pigs); diagnostic Neolithic features include stone cultivation/harvesting tools and textile technology. Clear material-culture parallels link Korea and western Japan.
• Genetics: Neolithic West Liao millet farmers carried substantial Amur-like ancestry with a temporal shift toward Yellow River ancestry; Early Neolithic West Liao genomes are lacking but Amur-like ancestry likely characterized early farmers. Neolithic individuals from Mongolia show high Amur-like ancestry, with increasing western Eurasian gene flow from the Bronze Age onward. Neolithic Koreans exhibit heterogeneous Jomon-related ancestry (0–95%) that declines over time, with present-day Koreans showing negligible Jomon contribution; Bronze Age Taejungni lacks significant Jomon component, consistent with migration into Korea associated with rice farming and modeled by Upper Xiajiadian-like ancestry. Yayoi farmers in Japan are modeled as Jomon plus Upper Xiajiadian-like admixture, supporting substantial Bronze Age migration from Korea into Japan. Ancient genomes from Nagabaka (southern Ryukyus) indicate Jomon-related origins with later turnover to Yayoi-like ancestry, contradicting a Holocene northward expansion from Taiwan.
• Integrated model: Two major phases characterize the spread: (1) Early–Middle Neolithic expansion of millet farmers with Amur-related ancestry from West Liao to adjacent regions, corresponding to primary linguistic splits; (2) Late Neolithic to Bronze/Iron Age admixture with Yellow River, western Eurasian, and Jomon populations, addition of rice and pastoralism, and extensive borrowing among daughter branches. Findings support the farming/language dispersal hypothesis over the pastoralist hypothesis for Transeurasian.

The converging linguistic, archaeological, and genetic evidence indicates that Transeurasian languages originated with Early Neolithic millet farmers in the West Liao region. Linguistic time depth and reconstructed agricultural vocabulary align with the onset and spread of millet cultivation and with archaeological clustering that traces millet’s dispersal to Korea and the Amur–Primorye by mid–6th millennium BP, and later incorporation of rice, wheat, and barley during the Bronze Age. Ancient DNA demonstrates predominant Amur-like ancestry in early farmers with later admixture from Yellow River, western Eurasian, and Jomon sources, mirroring the two-phase dispersal inferred linguistically and archaeologically. This integrated framework resolves the debate between pastoralist and farming models by showing that primary language splits correlate with Neolithic farming expansions, whereas Bronze Age interactions account for shared agropastoral vocabulary via borrowing rather than common inheritance. The results also situate Transeurasian origins alongside a multicentric model of millet domestication and parallel the association of Sino-Tibetan with Yellow River Neolithic farmers, underscoring regionally distinct centers of agricultural and linguistic expansions in Northeast Asia.

By triangulating large-scale linguistic datasets, an extensive archaeological site and crop database, and new and published ancient genomes, the study demonstrates that the early spread and primary diversification of Transeurasian languages were driven by Neolithic millet agriculture originating in the West Liao River region. Subsequent Bronze/Iron Age cultural contacts and gene flow introduced rice, western Eurasian crops, and pastoralism, generating widespread lexical borrowing among daughter branches. The integrated evidence overturns steppe-centered pastoralist models for primary dispersals and establishes agriculture as the principal driver of early Transeurasian language spread.

Key constraints include the absence of Early Neolithic genomes from the West Liao River region, limiting direct characterization of the earliest farmer ancestry there, and limited sample sizes/coverage for some ancient Korean individuals, which reduces resolution to test specific migration/replacement scenarios. Additionally, archaeological cultural groupings do not map one-to-one onto population movements, complicating direct attribution of migrations to single material cultures.