Results Predicting the density of human populations • Stepwise linear regression with population density as dependent variable → The density of population can be predicted with a high degree of accuracy (R > 0.8 ) by ecological factors Predicting the number of languages in a tile and language area • Stepwise linear regression with number of languages as dependent variable • Stepwise linear regression with language area as dependent variable → The number of languages in a cell and their area can be predicted with a good degree of accuracy by ecological factors (R > 0.75 & R > 0.8 resp.) •The number of languages at the putative origin of Bantu migrations is higher than what the regression predicts: density not fully explained by ecology? •The number of speakers can be predicted with a fair degree of accuracy by ecological factors (R =0.582) ( not shown ) Conclusions & Perspectives •Tight relationships between ecological factors and social/sociolinguistic va- riables → Given ‘layered’ paleo-environmental data, possibility to reconstruct migra- tory paths (assuming a time-enduring relationship to ecological contexts) • Deine better measure of l inguistic diversity, e.g. including the level of families and/or typological features •Better assess the putative origin of Bantu migrations References 1. Amante, C. & Eakins, B. W. (2009). ETOPO1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis. NOAA Technical Memorandum NESDIS NGDC-24. 2. Center for International Earth Science Information Network (CIESIN), Columbia University; and Centro Internacional de Agricultura Tropical (CIAT). 2005. Gridded Population of the World Ver- sion 3 (GPWv3): Population Density Grids. Palisades, NY: Socioeconomic Data and Applications Center (SEDAC), Columbia University. Available at http://sedac.ciesin.columbia.edu/gpw. (2011). 3. Ehret, C. (2001). Bantu Expansions: Re-Envisioning a Central Problem of Early African History. The Int. Journal of African Historical Studies 34(1): 5-41. 4. Food and Agriculture Organization & International Institute for Applied Systems Analysis (2000). Length of growing period, 1901-1996. Global agro-ecological zones. In FAO & IIASA, 2007, Map- ping biophysical factors that inluence agricultural production and rural vulnerability (H. von Velthuizen et al.) 5. Food and Agriculture Organization & International Institute for Applied Systems Analysis (2000). Coeficient of variation of length of growing period, 1901-1996. Global agro-ecological zones. In FAO & IIASA, 2007, Mapping biophysical factors that inluence agricultural production and rural vulnerability (H. von Velthuizen et al.) 6. GEODAS Grid Translator - Design-a-Grid. http://www.ngdc.noaa.gov/mgg/gdas/gd_designagrid.html 7. Global Mapping International & SIL International (2012). World Language Mapping System. Language area and point data for Geographic Information Systems (GIS). http://www.worldgeodatasets.com/language/ 8. Hansen, M., R. DeFries, J.R. Townshend, M. Carroll, C. Dimiceli, and R. Sohlberg (2003), Vegetation Continuous Fields MOD44B, 2001 Percent Tree Cover, Collection 3, University of Maryland, College Park, Maryland, 2001. 9. Jacquesson, F. (2001). Pour une linguistique des quasi-déserts. In A.M. Lofler-Laurian (ed.), Etude de linguistique générale et contrastive. Hommage à Jean Perrot. Paris : Centre de Recherche sur les Langues et les Sociétés, pp. 199-216. 10. Kummu, M., de Moel, H., Ward, P. J., Varis, O. (2011) How Close Do We Live to Water? A Global Analysis of Population Distance to Freshwater Bodies. PLoS ONE 6(6): e20578. doi:10.1371/journal. pone.0020578 11. Lewis, M. P. (ed.), 2009. Ethnologue: Languages of the World, Sixteenth edition. Dallas, Tex.: SIL International. http://www.ethnologue.com/. 12. Nettle, D. (1996). Language diversity in West Africa: An ecological approach. Journal of Anthropological Archaeolog y 15:403-438. 13. Nettle, D. (1998). Explaining global patterns of language diversity. Journal of Anthropological Archaeolog y 17: 354-74.. 14. Schwartz, D. (1992). Assèchement climatique vers 3 000 B.P. et expansion Bantu en Afrique centrale atlantique : quelques rélexions. Bulletin de la Société géologique de France 163(3) : 353-361. 15. Vansina, J. (1995). New Linguistic Evidence and ‘the Bantu expansion’. The Journal of African History 36(2): 173-195. Impact of climatic and ecological contexts on sociolinguistic factors among Bantu populations Christophe Coupé 1,2 & Jean-Marie Hombert 1 1 Laboratoire Dynamique du Langage, CNRS - Université de Lyon, 2 Institut Rhône-Alpin des systèmes complexes Background •The expansion of Bantu farmers 3,000 years ago may be linked to climatic changes during the Holocene period (Schwartz, 1992) •Questions remain about where Bantu populations came from and how they spread, especially when they faced the equatorial forest (Vansina, 1992) • Geographic variations in terms of diversity of the Bantu languages; does highest linguitic density mean point of origin (Ehret, 2001)? • Previous studies linking linguistic diversity to ecological factors, e.g. the length of the growing season (Nettle, 1996; Nettle, 1998; Jacquesson, 2001) Goal • To better understand the relationship between i) social and sociolinguistic fac- tors and ii) African ecological contexts •To later use this relationship to derive past human migrations from descrip- tions of their ecological contexts (e.g. with paleoclimatic models) Methodology and tools Methodology •To assemble a number of meaningful variables - elevation, vegetation etc. - describing African ecological contexts with high spatial resolution •To conduct statistical analyses of the correlations between these variables and i) population density, ii) linguistic diversity Datasets Statistical analyses • Tools: Quantum GIS , Excel, R & SPSS 19 •Statistical units: 368 cells covering African lands (from 10,000 to 90,000 km²) • Boxplot transformations for elv., rug., dist. to water, pop. dens. & ling. variables •Signi icant col inearity between the ecological variables → dificult to identify ‘primary’ vs. ‘secondary’ causal factors Contact: Christophe COUPE, christophe.coupe@ish-lyon.cnrs.fr Financial support: ANR CLHASS; ASLAN Laboratory of Excellence Elevation & rugosity % of herbaceous plants and trees Distance to fresh water Languages area & number of speakers Length of the growing season (LGP) Inter-annual CV of the LGP Population counts / density Units of statistical analysis Model 7: Coeficients Unstandardized Coeficients Standardized Coeficients B Std. Error Beta t sig (Constant) 3.638 2.260 1.610 .108 Distance to water -1.788 .345 -.383 -5.179 .000 % herbaceous .080 .010 1.054 8.359 .000 Elevation -.074 .009 -.301 -8.312 .000 Rugosity .500 .085 .249 5.886 .000 CV LGP .677 .155 .194 4.379 .000 % bare .073 .012 1.281 6.188 .000 LGP .342 .064 .593 5.306 .000 Model R R² Adjusted R² Std. Error of the Estimate 1 (a) .670 .449 .447 1.8012232 2 (b) .718 .515 .512 1.6919270 3 (c) .739 .547 .543 1.6378955 4 (d) .785 .616 .612 1.5101667 5 (e) .795 .632 .627 1.4802679 6 (f) .801 .641 .635 1.4633308 7 (g) .817 .667 .661 1.4112213 a. Predictors: (Constant), Distance to water b. Predictors: (Constant), Distance to water, % herbaceous c. Predictors: (Constant), Distance to water, % herbaceous, Elevation d. Predictors: (Constant), Distance to water, % herbaceous, Elevation, Rugosity e. Predictors: (Constant), Distance to water, % herbaceous, Elevation, Rugosity, CV LGP f. Predictors: (Constant), Distance to water, % herbaceous, Elevation, Rugosity, CV LGP, % bare g. Predictors: (Constant), Distance to water, % herbaceous, Elevation, Rugosity, CV LGP, % bare, LGP Model R R² Adjusted R² Std. Error of the Estimate 1 (a) .723 .523 .522 .6903445 2 (b) .734 .538 .536 .6799944 3 (c) .741 .550 .546 .6725704 4 (d) .746 .556 .551 .6687364 5 (e) .750 .562 .556 .6648291 6 (f) .753 .567 .560 .6621412 a. Predictors: (Constant), % bare b. Predictors: (Constant), % bare, Rugosity c. Predictors: (Constant), % bare, Rugosity, Pop density d. Predictors: (Constant), % bare, Rugosity, Pop density, CV LGP e. Predictors: (Constant), % bare, Rugosity, Pop density, CV LGP, Elevation f. Predictors: (Constant), % bare, Rugosity, Pop density, CV LGP, Elevation, LGP Model 6: Coeficients Unstandardized Coeficients Standardized Coeficients B Std. Error Beta t sig (Constant) 2.652 .299 8.863 .000 % bare -.013 .002 -.561 -6.071 .000 Rugosity -.216 .040 -.261 -5.438 .000 Pop density .094 .022 .229 4.254 .000 CV LGP -.145 .073 -.101 -1.983 .048 Elevation .012 .005 .119 2.658 .008 LGP .036 .018 .153 1.986 .048 Model R R² Adjusted R² Std. Error of the Estimate 1 (a) .786 .619 .618 1.9980498 2 (b) .802 .643 .641 1.9356130 3 (c) .809 .655 .652 1.9062422 4 (d) .819 .670 .666 1.8659055 5 (e) .821 .674 .670 1.8560117 a. Predictors: (Constant), % bare b. Predictors: (Constant), % bare, LGP c. Predictors: (Constant), % bare, LGP, Pop density d. Predictors: (Constant), % bare, LGP, Pop density, CV LGP e. Predictors: (Constant), % bare, LGP, Pop density, CV LGP, Elevation Model 5: Coeficients Unstandardized Coeficients Standardized Coeficients B Std. Error Beta t sig (Constant) 13.893 .834 16.649 .000 % bare .048 .006 .633 8.160 .000 LGP -.154 .049 -.201 -3.139 .002 Pop density -.218 .055 -.163 -3.943 .000 CV LGP .794 .203 .171 3.909 .000 Elevation .023 .010 .070 2.209 .028 L YON DE U NIVERSIT Institut des systmes complexes Complex Systems Institute Rhne-Alpes IXXI