Kinship structures create persistent channels for language transmission

Abstract

Languages are transmitted through channels created by kinship systems. Given sufficient time, these kinship channels can change the genetic and linguistic structure of populations. In traditional societies of eastern Indonesia, finely resolved cophylogenies of languages and genes reveal persistent movements between stable speech communities facilitated by kinship rules. When multiple languages are present in a region and postmarital residence rules encourage sustained directional movement between speech communities, then languages should be channeled along uniparental lines. We find strong evidence for this pattern in 982 individuals from 25 villages on two adjacent islands, where different kinship rules have been followed. Core groups of close relatives have stayed together for generations, while remaining in contact with, and marrying into, surrounding groups. Over time, these kinship systems shaped their gene and language phylogenies: Consistently following a postmarital residence rule turned social communities into speech communities.

Keywords: coevolution; cultural evolution; kinship; language; population genetics.

Fig. 1.

Languages in sampled villages on Sumba and Timor, eastern Indonesia. Pie charts show the languages spoken, scaled by sample size. (Left) The 14 patrilocal villages (■) on Sumba and the Austronesian languages spoken by the 505 sampled men. (Right) The 11 communities on Timor, including 9 matrilocal villages (•) and 2 patrilocal villages (■). Each of the 477 men sampled on Timor speaks one or more of five local languages belonging to two language families, Austronesian (Dawanta, Kemak, Betun, and Upper Tetun) and non-Austronesian Bunak.

Fig. 2.

Language sharing in the phylogenies of Sumba (A and B) and Timor (C and D). (A and C) mtDNA. (B and D) Y chromosome. Color bands beneath the phylogenies show the languages spoken by each individual (monolingual on Sumba; sometimes multilingual on Timor). Plots to the right of each phylogeny show the probability of sharing a language $l$ given that each pair of individuals are in the same genetic clade $g$ at a given time in the past. Solid lines represent the observed metric, with shaded bands indicating the result of random permutations of the linguistic data. Higher probabilities that close genetic relatives share a language, compared with random expectations, were observed for Sumba Y (B) and Timor mtDNA (C) at all time periods.

Fig. 3.

Genetic structure from an IM model with migration influenced by kinship practices. (A–D) The theoretical role of kinship practices on genetic diversity is shown for populations that are endogamous (A), ambilocal or neolocal (B), patrilocal (C), and matrilocal (D). (E) Close correspondence of the IM model (red shading) with observed data (blue contours) for patrilocal Sumba using ($N=144$, $n=69$, $m_{\mathit{female}}=0.21$, $m_{\text{male}}=0.01$, $\tau =7.61$, $a=3.31$). (F) Close correspondence of the IM model with only matrilocal villages on Timor, using ($N=81$, $n=76$, $m_{\mathit{female}}=0.12$, $m_{\text{male}}=0.19$, $\tau =23.65$, $a=1.31$). Insets in E and F show the posterior distributions of migration rates for the Sumba and Timor kinship systems based on 3 million samples drawn from prior distributions of all IM model parameters.

Fig. 4.

Genetic distances on Sumba (A and C) and Timor (B and D), between all pairs of individuals (A and B) and only between individuals who speak a common language (C and D). Conditioning on language sharing reveals three distinct clusters in the Sumba data (C), showing evidence of village endogamy (e), ambilocality or neolocality (an), and patrilocality (p). For Timor, the high degree of multilinguality means that most pairs of individuals in B are also included in D. B and D are dominated by matrilocality (m) and ambilocality or neolocality (an), with a small patrilocal cluster (p).

Fig. 5.

The Z score measure of association between gene and language phylogenies on Sumba and Timor for different language switching rates $\alpha$. Language switching rates below ∼0.5% per generation are required to generate the type of association between languages and clades observed in the empirical data (Table 1). All cases independently converge and abruptly lose gene–language associations, behaving similarly to randomized cases, when the language switch rate exceeds ∼0.5% per generation. Inset zooms in to show the fine scale of inferred host-switching rates.