Phonemic Diversity Supports a Serial Founder Effect Model of

Language Expansion from Africa

Quentin D. Atkinson
Science 332, 346 (2011);
DOI: 10.1126/science.1199295

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your
colleagues, clients, or customers by clicking here.
Permission to republish or repurpose articles or portions of articles can be obtained by
following the guidelines here.

The following resources related to this article are available online at

[Link] (this infomation is current as of April 17, 2011 ):

Downloaded from [Link] on April 18, 2011

Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
[Link]
Supporting Online Material can be found at:
[Link]
This article cites 23 articles, 10 of which can be accessed free:
[Link]
This article appears in the following subject collections:
Evolution
[Link]

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2011 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.
REPORTS
ability to control the intricate structure of DNA 12. C. Zhang et al., Proc. Natl. Acad. Sci. U.S.A. 105, Research, Army Research Office, National Science
nano-architectures and create more diverse build- 10665 (2008). Foundation, Department of Energy, and National
13. C. Lin, Y. Liu, H. Yan, Biochemistry 48, Institutes of Health to H.Y. and Y.L. and from the
ing blocks for molecular engineering. 1663 (2009). Sloan Research Foundation to H.Y. Y.L. and H.Y. were
14. P. W. K. Rothemund, Nature 440, 297 (2006). supported by the Technology and Research Initiative
References and Notes 15. E. S. Andersen et al., Nature 459, 73 (2009). Fund from Arizona State University and as part of the
1. N. C. Seeman, J. Theor. Biol. 99, 237 (1982). 16. S. M. Douglas et al., Nature 459, 414 (2009). Center for Bio-Inspired Solar Fuel Production, an
2. N. C. Seeman, Nature 421, 427 (2003). 17. Y. Ke et al., J. Am. Chem. Soc. 131, 15903 (2009). Energy Frontier Research Center funded by the U.S.
3. T. J. Fu, N. C. Seeman, Biochemistry 32, 3211 (1993). 18. H. Dietz, S. M. Douglas, W. M. Shih, Science 325, Department of Energy, Office of Science, Office of
4. J. H. Chen, N. C. Seeman, Nature 350, 631 (1991). 725 (2009). Basic Energy Sciences under award DE-SC0001016.
5. E. Winfree, F. Liu, L. A. Wenzler, N. C. Seeman, 19. Materials and methods are available as supporting
Nature 394, 539 (1998). material on Science Online.
6. H. Yan, S. H. Park, G. Finkelstein, J. H. Reif, T. H. LaBean, 20. P. W. K. Rothemund et al., J. Am. Chem. Soc. 126,
Supporting Online Material
16344 (2004). [Link]/cgi/content/full/332/6027/342/DC1
Science 301, 1882 (2003).
Materials and Methods
7. W. M. Shih, J. D. Quispe, G. F. Joyce, Nature 427, 618 21. P. J. Hagerman, Annu. Rev. Biophys. Chem. 17, 265
SOM Text
(2004). (1988).
Figs. S1 to S39
8. F. Mathieu et al., Nano Lett. 5, 661 (2005). Acknowledgments: We thank the AFM applications group
Tables S1 to S19
9. R. P. Goodman et al., Science 310, 1661 (2005). at Bruker Nanosurfaces for assistance in acquiring some
Reference 22
10. F. A. Aldaye, H. F. Sleiman, J. Am. Chem. Soc. 129, of the high-resolution AFM images, using Peak Force
13376 (2007). Tapping with ScanAsyst on the MultiMode 8. This research 18 January 2011; accepted 4 March 2011
11. Y. He et al., Nature 452, 198 (2008). was partly supported by grants from the Office of Naval 10.1126/science.1202998

Downloaded from [Link] on April 18, 2011

0.131, df = 503, P = 0.003). To account for any
Phonemic Diversity Supports a Serial non-independence within language families, the
analysis was repeated, first using mean values at
Founder Effect Model of Language the language family level (table S2) and then
using a hierarchical linear regression framework
Expansion from Africa to model nested dependencies in variation at the
family, subfamily, and genus levels (15). These
analyses confirm that, consistent with a founder
Quentin D. Atkinson1,2* effect model, smaller population size predicts
reduced phoneme inventory size both between
Human genetic and phenotypic diversity declines with distance from Africa, as predicted by a families (family-level analysis r = 0.468, df = 49,
serial founder effect in which successive population bottlenecks during range expansion P < 0.001; fig. S1B) and within families, con-
progressively reduce diversity, underpinning support for an African origin of modern humans. trolling for taxonomic affiliation {hierarchical lin-
Recent work suggests that a similar founder effect may operate on human culture and ear model: fixed-effect coefficient (b) = 0.0338
language. Here I show that the number of phonemes used in a global sample of 504 languages to 0.0985 [95% highest posterior density (HPD)],
is also clinal and fits a serial founder–effect model of expansion from an inferred origin in P = 0.009}.
Africa. This result, which is not explained by more recent demographic history, local language Figure 1B shows clear regional differences in
diversity, or statistical non-independence within language families, points to parallel phonemic diversity, with the largest phoneme
mechanisms shaping genetic and linguistic diversity and supports an African origin of inventories in Africa and the smallest in South
modern human languages. America and Oceania. A series of linear regres-
sions was used to predict phoneme inventory size
he number of phonemes—perceptually ing the evolution of phonemes (11, 16) and lan- from the log of speaker population size and dis-

T distinct units of sound that differentiate

words—in a language is positively corre-
lated with the size of its speaker population (1) in
guage generally (17–20). This raises the possibility
that the serial founder–effect model used to trace
our genetic origins to a recent expansion from Africa
tance from 2560 potential origin locations around
the world (15). Incorporating modern speaker
population size into the model controls for geo-
such a way that small populations have fewer (4–9) could also be applied to global phonemic graphic patterning in population size and means
phonemes. Languages continually gain and lose diversity to investigate the origin and expansion that the analysis is conservative about the amount
phonemes because of stochastic processes (2, 3). of modern human languages. Here I examine geo- of variation attributed to ancient demography.
If phoneme distinctions are more likely to be lost graphic variation in phoneme inventory size using Model fit was evaluated with the Bayesian in-
in small founder populations, then a succession data on vowel, consonant, and tone inventories formation criterion (BIC) (22). Following previ-
of founder events during range expansion should taken from 504 languages in the World Atlas of ous work (5, 6), the set of origin locations within
progressively reduce phonemic diversity with in- Language Structures (WALS) (21), together with four BIC units of the best-fit location was taken
creasing distance from the point of origin, paral- information on language location, taxonomic to be the most likely area of origin under a serial
leling the serial founder effect observed in population affiliation, and speaker demography (Fig. 1 and founder–effect model.
genetics (4–9). A founder effect has already been table S1) (15). The origin locations producing the strongest
used to explain patterns of variation in other cul- Consistent with previous work (1), speaker decline in phonemic diversity and best-fit model
tural replicators, including human material culture population size is a significant predictor of phone- lie across central and southern Africa (Fig. 2A).
(10–13) and birdsong (14). A range of possible mic diversity (Pearson’s correlation r = 0.385, This region could represent either a single origin
mechanisms (15) predicts similar dynamics govern- df = 503, P < 0.001), with smaller population size for modern languages or the main origin under
predicting smaller overall phoneme inventories a polygenesis scenario. The best-fit model in-
1 (fig. S1A). The same relationship holds for vow- corporating population size and distance from
Department of Psychology, University of Auckland, Private
Bag 92019, Auckland, New Zealand. 2Institute of Cognitive and el (r = 0.378, df = 503, P < 0.001) and tone (r = the origin explains 31% of the variance in pho-
Evolutionary Anthropology, University of Oxford, 64 Banbury 0.230, df = 503, P < 0.001) inventories sepa- neme inventory size [correlation coefficient (R) =
Road, Oxford OX2 6PN, UK. rately,with a weaker,though still significant, effect 0.558, F2,501 = 113.463, P < 0.001] (Fig. 3). Both
*E-mail: [Link]@[Link] of population size on consonant diversity (r = population size (rpopulation = 0.146, P = 0.002)

346 15 APRIL 2011 VOL 332 SCIENCE [Link]

REPORTS

Fig. 1. Language locations and regional variation in phonemic diversity. (A) Map region reveal substantial regional variation (c2 = 188.7, df = 5, P < 0.001), with the
showing the location of the 504 sampled languages for which phoneme data was highest diversity in Africa and the lowest diversity in Oceania and South America.

Downloaded from [Link] on April 18, 2011

compiled from the WALS database. (B) Box plots of overall phonemic diversity by The same regional pattern also applies at the language family level (fig. S2).

and distance from origin (rdistance = –0.438, P <

0.001) are significant predictors in the model.
Controlling for population size, distance from
origin accounts for 19% of the variance in pho-
nemic diversity. A model using only distance as a
predictor gives a broadly equivalent origin area
(fig. S3) and explains 30% of the variation in
phonemic diversity (r = 0.545, P < 0.001). The
relationship also holds for vowel (r = –0.394, P <
0.001), consonant (r = –0.260, P < 0.001), and
tone diversity (r = –0.391, P < 0.001) separately.
To account for relatedness within families, I
repeated the above regressions using mean values
across language families (table S2) and under a
hierarchical linear model comprising the three
taxonomic levels recorded in WALS (15). The
hierarchical model results closely matched those
of the individual language analysis (fig. S4). Add-
ing an interaction effect did not significantly im-
prove model fit, indicating that the patterns
reported here reflect a consistent trend that holds
across the globe. The family-level analysis was
consistent with the individual language analyses,
although the credible region of origin is expanded
to include all of Africa (Fig. 2B). Distance from
the best fit origin (rdistance = – 0.401, P = 0.004)
and population size (rpopulation = 0.300, P =
0.036) are both significant predictors and account
for 39% of the variance in phonemic diversity
Fig. 2. Likely area of language origin. Maps show the likely location of a single language origin
between families (R = 0.627, F2,47 = 15.190, P <
under a founder effect model of phonemic diversity (controlling for population size) inferred
0.001; fig. S5). As a further test of the robustness
from (A) individual languages and (B) mean diversity across language families. Lighter shading
of these findings, individual regressions were implies a stronger inverse relationship between phonemic diversity and distance from the origin
repeated using partial Mantel tests, which allow and better fit of the model, as measured by the BIC. The most likely region of origin, comprising
for non-independence between data points and those locations within four BIC units of the best-fit origin location, is the area of lightest shading
avoid assumptions about the statistical distribu- outlined in bold.
tions underlying the variables of interest (15).
The results of this analysis matched the findings
reported above (table S3). ysis did not show a significant negative corre- does not appear under the hierarchical linear model
To examine the possibility of language poly- lation between distance from a second origin and or language family–level analysis; adding a sec-
genesis, distance from a second origin location phonemic diversity. Restricting the analysis to ond origin does not improve the fit of either model
was added as a predictor to a model incorporating second origin locations that do show an inverse as measured by the BIC, and all putative second
population size and distance from the best-fit relationship, a region of best fit can be identified origin locations are within four BIC units. The
origin in Africa. The best-fit models in this anal- in South America (fig. S6). However, this pattern area identified in the individual-level analysis may

[Link] SCIENCE VOL 332 15 APRIL 2011 347

REPORTS
Fig. 3. Phonemic diversity versus tion growth immediately after the African exodus,
distance from the best-fit origin in meaning that languages in this region should
Africa. A plot of distance from the have been least affected by population bottle-
best-fit origin location in Africa necks and would have had the most time to re-
against overall phoneme inventory cover diversity.
size is shown. Distance from the Although distance from Africa explains much
origin alone explains 30% of the less of the variation in phonemic diversity (19%)
variation in phonemic diversity than in neutral genetic markers (80 to 85%) (5, 7),
(fitted line; r = –0.545, n = 504 the effect is comparable to that obtained from
languages, P < 0.001) and 19.2% analysis of human mitochondrial DNA (18%) (8)
of the variation after controlling for
or phenotypic data (14 to 28%) (6). To the extent
modern speaker population size
that language can be taken as an example of
(rdistance = –0.438, P < 0.001;
rpopulation = 0.146, P < 0.001; R = cultural evolution more generally, these findings
0.558, F2,501 = 113.463, P < support the proposal that a cultural founder effect
0.001). operated during our colonization of the globe,
potentially limiting the size and cultural com-
plexity of societies at the vanguard of the human
expansion (10, 11). An origin of modern lan-
guages predating the African exodus 50,000 to

Downloaded from [Link] on April 18, 2011

be an artefact of slightly higher diversity levels in unit of area), population density (the number of 70,000 years ago puts complex language along-
the Americas than in Oceania, despite the two speakers per unit of area), or language area (the side the earliest archaeological evidence of sym-
being comparable distances from Africa, possibly total area over which a language is spoken) could bolic culture in Africa 80,000 to 160,000 years
because of a stronger founder effect across the affect regional phonemic diversity by increasing ago (27, 28). Truly modern language, akin to
remote Pacific (18–20, 23). When the languages contact within and between groups and creating languages spoken today, may thus have been the
of Oceania are removed from the individual anal- more opportunities to borrow new phonemes. To key cultural innovation that allowed the emer-
ysis, the effect of distance from an African origin test whether any such effect could explain the gence of these and other hallmarks of behavioral
remains (rdistance1 = –0.447, P < 0.001) but there is observed global cline in phonemic diversity, modernity and ultimately led to our colonization
no significant effect of distance from the second- these additional measures were included, togeth- of the globe (29).
ary origin (rdistance2 = –0.065, P = 0.192). As ex- er with population size and distance from the
pected if language spread south with the initial best-fit origin in Africa, in a regression model References and Notes
colonization of the Americas, distance from the predicting phoneme inventory size (15). Control- 1. J. Hay, L. Bauer, Language 83, 388 (2007).
Bering Strait is inversely correlated with pho- ling for other demographic variables in this way, 2. C. D. Yang, Knowledge and Learning in Natural
Language (Oxford Univ. Press, Oxford, 2003).
neme inventory size within the Americas after sub-Saharan Africa remains the most likely area
3. W. Labov, Principles of Linguistic Change: Social Factors
controlling for population size (rdistance = –0.173, of origin (fig. S7). Distance from the best-fit origin (Blackwell, Oxford, UK, 2001).
P = 0.043). location is a significant predictor at the individual 4. S. Ramachandran et al., Proc. Natl. Acad. Sci. U.S.A.
An ostensibly global cline in phonemic di- language level (rdistance = –0.413, P < 0.001), 102, 15942 (2005).
versity supporting an expansion from Africa could family level (rdistance = –0.384, P < 0.008), and in 5. F. Prugnolle, A. Manica, F. Balloux, Curr. Biol. 15,
R159 (2005).
also arise as an artefact of a series of more recent the hierarchical linear model [b = –3.419 × 10−5 to 6. A. Manica, W. Amos, F. Balloux, T. Hanihara, Nature 448,
expansions after the Last Glacial Maximum (LGM) –2.223 × 10−5 (95% HPD), P < 0.001]. The demo- 346 (2007).
into northern Eurasia, the Americas, and the re- graphic variables are highly correlated and did not 7. J. Z. Li et al., Science 319, 1100 (2008).
mote Pacific. Languages in these regions show show significant independent effects on phonemic 8. F. Balloux, L. J. L. Handley, T. Jombart, H. Liu, A. Manica,
Proc. Biol. Sci. 276, 3447 (2009).
lower average phonemic diversity than in the rest diversity. Stepwise regressions indicated that a 9. L. Betti, F. Balloux, W. Amos, T. Hanihara, A. Manica,
of the world (t = –6.597, df = 503, P < 0.001), model incorporating distance from Africa, popu- Proc. Biol. Sci. 276, 809 (2009).
which is consistent with a more recent coloniza- lation size, and (at the individual language level 10. J. Diamond, Nature 273, 185 (1978).
tion. However, expansion after the LGM does only) language area best explained phoneme in- 11. J. Henrich, Am. Antiq. 69, 197 (2004).
12. S. J. Lycett, N. von Cramon-Taubadel, J. Archaeol. Sci. 35,
not account for the global cline in phonemic ventory size (15). 553 (2008).
diversity. Distance from Africa remains a signif- The single major cline in phonemic diver- 13. D. S. Rogers, M. W. Feldman, P. R. Ehrlich, Proc. Biol. Sci.
icant predictor of phonemic diversity after con- sity is consistent with a linguistic founder effect 276, 3835 (2009).
trolling for colonization since the LGM (rdistance = operating under conditions of rapid expansion 14. A. J. Baker, P. F. Jenkins, Anim. Behav. 35, 1793
(1987).
–0.401, P < 0.001; rpopulation = 0.152, P = 0.001; from a most likely origin in Africa. This sup-
15. Materials and methods are available as supporting
rLGM = 0.032, P = 0.419), as well as when these ports a picture of language spread that is con- material on Science Online.
more recently colonized areas are excluded from gruent with similar analyses of human genetic 16. B. de Boer, J. Phonetics 28, 441 (2000).
the analysis altogether (rdistance = –0.511, P < (4, 5, 7, 8) and phenotypic (6, 9) diversity. Pho- 17. L. L. Cavalli-Sforza, M. W. Feldman, Cultural Transmission
0.001; rpopulation = 0.253, P < 0.001) (19). nemic diversity appears to be highly stable with- and Evolution: A Quantitative Approach (Princeton Univ.
Press, Princeton, NJ, 1981).
Demographic factors other than population in major language families (15), indicating that, 18. Q. D. Atkinson, R. D. Gray, Syst. Biol. 54, 513 (2005).
size may also influence phonemic diversity, par- despite the many sociolinguistic processes at 19. Q. D. Atkinson, A. Meade, C. Venditti, S. J. Greenhill,
ticularly those affecting levels of contact and work (2, 3), robust statistical patterns in global M. Pagel, Science 319, 588 (2008).
borrowing between groups of speakers. Because variation can persist for many millennia and could 20. P. Trudgill, Linguist. Typol. 8, 305 (2004).
21. M. Haspelmath, M. S. Dryer, D. Gil, B. Comrie, Eds.,
neighboring populations at similar points in an plausibly reflect a time scale on the order of The World Atlas of Language Structures Online (Max Planck
expansion are more likely to have similar pho- the African exodus. Outside Africa, the highest Digital Library, Munich, 2008).
nemes and levels of diversity, moderate horizon- levels of phonemic diversity are found in lan- 22. K. P. Burnham, D. R. Anderson, Model Selection and
tal transfer between populations can maintain a guage families thought to be autochthonous to Inferences (Springer, New York, 1998).
23. P. V. Kirch, R. C. Green, Curr. Anthropol. 28, 431
cline, as has been the case for human genetic Southeast Asia. This also fits with genetic evi- (1987).
diversity (24). However, geographic variation in dence (25, 26), indicating that Southeast Asia 24. L. J. L. Handley, A. Manica, J. Goudet, F. Balloux,
language diversity (the number of languages per experienced particularly pronounced popula- Trends Genet. 23, 432 (2007).

348 15 APRIL 2011 VOL 332 SCIENCE [Link]

REPORTS
25. M. G. Palanichamy et al., Am. J. Hum. Genet. 75, 966 30. Thanks to M. Pagel, O. Curry, R. Dunbar, M. Dunn, SOM Text
(2004). R. Gray, S. Greenhill, M. Grove, S. Roberts, R. Ross, and Figs. S1 to S8
26. Q. D. Atkinson, R. D. Gray, A. J. Drummond, Mol. Biol. S. Schultz for useful advice and/or comments on the Tables S1 to S4
Evol. 25, 468 (2008). manuscript. I declare no competing financial interest. References
27. C. Henshilwood, F. d’Errico, M. Vanhaeren, K. van Niekerk,
Z. Jacobs, Science 304, 404 (2004). Supporting Online Material
28. C. W. Marean et al., Nature 449, 905 (2007). [Link]/cgi/content/full/332/6027/346/DC1 19 October 2010; accepted 4 March 2011
29. I. Tattersall, Proc. Natl. Acad. Sci. U.S.A. 106, 16018. Materials and Methods 10.1126/science.1199295

resolution data on multiple forcing mechanisms

Interplay Between Changing Climate should be allied to paleontological, species-level
phylogenies constructed on sufficiently complete
and Species’ Ecology Drives fossil records over substantial periods of evolu-
tionary history. This resolution is rare (2, 6 ),
Macroevolutionary Dynamics but Cenozoic macroperforate planktonic forami-
nifera provide a suitable record for testing these
hypotheses (13).
Thomas H. G. Ezard,1,2* Tracy Aze,3 Paul N. Pearson,3 Andy Purvis1 Planktonic foraminifera are sexually repro-
ducing protists distributed throughout the world’s

Downloaded from [Link] on April 18, 2011

Ecological change provokes speciation and extinction, but our knowledge of the interplay among oceans. The calcium carbonate “shells” (known
the biotic and abiotic drivers of macroevolution remains limited. Using the unparalleled fossil as “tests”) of dead individuals rain down on
record of Cenozoic macroperforate planktonic foraminifera, we demonstrate that macroevolutionary the ocean floor and can, under favorable condi-
dynamics depend on the interaction between species’ ecology and the changing climate. This tions, generate continuous microfossil sequences
interplay drives diversification but differs between speciation probability and extinction risk: Speciation that span millions of years. The group’s useful-
was more strongly shaped by diversity dependence than by climate change, whereas the reverse ness for stratigraphic correlation (14) and paleo-
was true for extinction. Crucially, no single ecology was optimal in all environments, and species climatic reconstruction (15) has led to extensive
with distinct ecologies had significantly different probabilities of speciation and extinction. The documentation of its morphology (14) and depth
ensuing macroevolutionary dynamics depend fundamentally on the ecological structure of habitats (13). The phylogenetic relationships
species’ assemblages. within the macroperforate clade of the Cenozoic
have recently been revised comprehensively (13),
via the application of Simpson’s evolutionary
he wide-ranging mechanisms that gener- and therefore offer little insight into extinction species concept (16 ). Under this concept, each

T ate and maintain biodiversity have been

grouped in many different ways, but a
fundamental distinction exists between biotic and
(11, 12). To distinguish how interwoven biotic
and abiotic processes regulate diversity, high-
species is intended to represent a single line of
descent (16) that begins with a speciation (clado-

abiotic drivers (1–5). If interactions among species

50
are the dominant drivers of evolution, as in the A
“Red Queen” model (1, 4), then diversification
40
rates among groups of interacting species are
sub thermocline
expected to show diversity-dependent dynamics
with ongoing turnover at equilibrium. A conse- 30
quence of this mechanism is that diversification
rates are expected to decrease as a function of 20 thermocline
diversity. Conversely, if evolution is driven chief-
ly by changes in the physical environment—as 10
in the “Court Jester” model (3), named to con- mixed layer
Diversity

trast with the Red Queen—macroevolutionary 0

dynamics should be dominated by cladewide 50
B
effects of abrupt abiotic perturbations. Although
the interplay between these alternative drivers 40
has long been recognized as fundamental for keels
regulating diversity (6, 7), progress toward un- 30
derstanding their interaction has been slow (2). keels & spines &
The incompleteness of the fossil record often 20 symbionts symbionts
necessitates temporally and taxonomically coarse
spines
paleontological analyses (7–10), whereas molec- 10 symbionts
ular phylogenies are restricted to extant species others
0
1
Division of Biology, Silwood Park Campus, Imperial College
London, Ascot, Berkshire, SL5 7PY, UK. 2Department of Paleocene Eocene Oligocene Miocene Pl Pt
Mathematics, Faculty of Engineering and Physical Sciences,
University of Surrey, Guildford, Surrey, GU2 7XH, UK. 3School 65 60 50 40 30 20 10 0
of Earth and Ocean Sciences, Cardiff University, Cardiff, CF10
3YE, UK. Fig. 1. The relative frequencies of depth habitats (A) and morphologies (B) of macroperforate planktonic
*To whom correspondence should be addressed. E-mail: foraminifer species across the Cenozoic (13) has fluctuated substantially. Time (million years before the
[Link]@[Link] present) is based on the marine geological time scale (14). Pl, Pliocene; Pt, Pleistocene.

[Link] SCIENCE VOL 332 15 APRIL 2011 349