Skip to main content
. 2022 Dec 9;13(1):20220053. doi: 10.1098/rsfs.2022.0053

Figure 6.

Figure 6.

Low-dimensional representations of root (a) and formative-based (b) representations of body-part morphological structure. We applied PCA to the distance matrix M, to provide a two-dimensional representation using root and formative-based features. Root-based low-dimensional representations presents a single cluster (with a continuum-like organization regarding PCA 2 dimension) and two outliers: Kaxarari and Matses. Formative-based low-dimensional representations present one cluster, which randomly comprises languages from different branches, leaving the remaining languages in a radically discontinuous distribution.