Originally Posted by grant hutchison

Phonosemantics takes this a little farther, noting word clusters with common sounds and meanings, unrelated to simple onomatopoeia. One can exploit this with invented words, as in the famous "bouba/kiki" experiment. Subjects exposed to these words for the first time, and asked to associate them with objects, would reproducibly identify the rounded, lumpy object as the "bouba", and the spiky, sharp one as the "kiki". This happens with subjects who speak languages from many different language groups. Does it go back to the sounds of lumpy and sharp objects interacting with the environment? Maybe. But there are other examples that are purely visual, like the "gl-" group that associates with light sources (gleam, glitter, glister, glisten, glow, glint, in English).