1. Serial Founder Effects in Linguistics and Genetics Claire Bowern (with Keith Hunley and Meghan Healy) Yale and University of New Mexico Feb 9, 2012 Based on Hunley et al: (2012) Rejection of a Serial Founder Effects model… Roy Soc. Proc. B. 2/01

2. Genes and Languages  Consensus about broad-scale similarities in evolutionary processes in linguistics and genetics  Discrete heritable units (DNA sequences :: words, phonemes (= distinctive sounds in a language), grammar, etc)  … which undergo mutation  … at different rates.  Can identify homologous features which descend from common ancestors (with understanding of the processes of change, we can reconstruct features of those ancestors)  Transmission is both vertical and lateral, with the former (usually) predominating  Can be modeled by trees and networks

3. Genes and Languages  Change thought to operate at very different time scales:  >6000 distinct languages currently spoken, in about 150 families.  Much current attested diversity dates to within the last 10,000 years:  Indo-European, Uralic, Afro-Asiatic [inc Semitic], Sino-Tibetan  Austronesian, Pama-Nyungan, Algonquian  etc.  (Although language has been around for at least 100,000 years.)

4. Atkinson (2011)  Claim:  Language Shows Serial Founder Effects: “Human genetic and phenotypic diversity declines with distance from Africa, as predicted by a serial founder effect in which successive population bottlenecks during range expansion progressively reduce diversity, … Recent work suggests that a founder effect may operate on human culture and language. Here I show that the number of phonemes used in a global sample of 504 languages is also clinal and fits a serial founder– effect model of expansion from an inferred origin in Africa. This result … points to parallel mechanisms shaping genetic and linguistic diversity and supports an African origin of modern human languages.”

5. Background:  Hay and Bauer (2007) showed that there was a weak but significant correlation between number of speakers of a language and the number of distinctive sounds in a language, but could provide no reason for the correlation.  Atkinson uses the correlation to hypothesize a SFE process:  Correlation points to Founder Effect  Migrant populations exhibit a reduced amount of allele/phoneme diversity, which gradually recovers over time.  Greatest genetic/phonemic diversity found in Africa  Clinal decrease in genetic/phonemic diversity as distance from Africa increases.

6. Atkinson (2011):  If true:  Language retains information about prehistory for much longer than previously claimed.  Therefore important new source of information about global prehistory.  However:  Atkinson only tested one prediction of a SFE model.

7. Today:  Testing further empirical predictions of SFE models in genetic and linguistic samples.  Genetics:  614 unlinked, autosomal microsatellite loci  2,251 individuals in 100 populations.  Linguistics:  908 phonemes scored as ‘present/absent’  725 languages

8. Languages in Sample

9. Phonemes  “distinctive sounds” (sounds which combine to form words and whose substitution causes a change in meaning):  English:  pat ≠ bat (p and b are distinct phonemes)  p ʰ it ~ spit (p and p ʰ are not distinct phonemes; vary by position in word)  Bardi:  k ʰ a ː ra ~ ka ː ra ~ ga ː ra ‘uncle’ (k ʰ, k, and g are not distinct phonemes)  a ɹ a ‘other’ ≠ ara ‘no’ ( ɹ and r are distinct)

10. Four Predictions: 1. Africans will possess more unique alleles/phonemes than people from other regions (number of ‘private’ alleles/phonemes) 2. There will be a negative correlation between within-group variation and geographic distance from the African origin. 3. The pattern of among-group variation will be tree-like, and the trees will be rooted in Africa. 4. The degree of difference among groups will reflect the pattern of branching in the tree. Correlations between patterns of among-group variation and geographic distance will be purely a byproduct of the splitting and movement process, not exchange between neighboring groups.

11. 1: ‘Private’ Alleles/Phonemes More unique alleles/phonemes in Africa

12. Shared vs Unique Alleles/Phonemes  Many fewer phonemes than alleles shared across populations (implies greater rates of change in language)  More unique alleles/phonemes in Africa than non-Africa

13. Total vs Private Alleles/Phonemes (by region)

14. 2: Within group variation and geographic distance from Africa Reduced heterozygosity as distance from Africa increases

16. 3: Tree-like patterns, with tree rooted in Africa

20. 4: Degrees of difference between groups will reflect branching patterns in tree. [geography will play a role only as a by-product of splitting processes]

22. Afroasiatic and Africa: Phoneme distance vs Geographic distance

23. Conclusions

24. SFE not supported in linguistics  Discordant patterns between genetics and linguistics, with the genetic patterns all in agreement.  Non-tree-like patterns in phonemic variation.  (Phonemes not parallel to genes?)  some correlations between distance and regions [independent of language family] – implies that pattern is, at least in part, driven by local exchange (borrowing, language contact)  rates of change in language are too rapid to preserve early SFE.

25. Acknowledgements  NSF grant BCS-920114