Clustering in the creole formation process.

<p>Points are the projection of the census data (See Tables A-D in the <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0120771#pone.0120771.s001" target="_blank">S1 Supporting Information</a>) in the plane (<i>N</i>_<i>M</i>/(<i>N</i>_<i>B</i>+<i>N</i>_<i>M</i>), (<i>N</i>_<i>M</i>+<i>N</i>_<i>B</i>)/(<i>N</i>_<i>M</i>+<i>N</i>_<i>B</i>+<i>N</i>_<i>Eu</i>)). Red circles mark States where a creole language emerged while purple ones identify States where a creole language historically did not emerge. The gray stripe is the outcome of our modeling scheme and separates the regions where respectively the creole <i>C</i> (above the stripe) and the European <i>E</i> (below the stripe) represent the dominant language (i.e., spread among more than the 80% of the population) in the Mulattos and Bozal populations in the asymptotic states of the model. The two black curves delimiting the gray stripe and the dashed line in the middle are obtained by simulations performed with the same parameters <i>γ</i> = 0.8, <i>δ</i> = 0.1 and <i>N</i> = <i>N</i>_<i>M</i>+<i>N</i>_<i>B</i>+<i>N</i>_<i>Eu</i> = 10000, with <i>ε</i> ranging from 0.05 (bottom black curve) to 0.07 (upper black curve), passing through 0.06 (dashed curve). The horizontal axis has been artificially expanded by a power 0.2.