Figure 3.
Correlations between number of cognate sets (x-axis, NOS) and rate of lexical replacement, in (a) the IE languages for 200 Swadesh list meanings; (b) the Bantu languages for n = 102 meanings; (c) the Austronesian languages for n = 154 meanings. The darker circles correspond to the low-limit number words. The rate and number of states for one are relatively high in the Bantu and Austronesian datasets.