Which colexifications in CLICS are cross-linguistically universal versus areally or genealogically driven?

A colexification occurs when a language uses one word form for two distinct concepts (e.g. TREE/WOOD, SEE/KNOW). CLICS3 aggregates such patterns across thousands of languages into a network whose nodes are concepts and whose edge weights count colexifying languages. A high edge weight can reflect a genuine cross-linguistic (cognitive/semantic) universal or merely inheritance within one family or diffusion within one area. QUESTION: over CLICS3, for each colexified concept pair compute the number of distinct language families and macroareas that attest it, and test its recurrence against a genealogically-and-areally controlled null (permutation stratified by family and macroarea); produce a ranked list of control-surviving 'universal' colexifications with effect sizes, and determine which of the strongest raw CLICS edges are actually family/area confounded. Additionally, characterize whether the global colexification network's degree distribution and community structure are stable under the control.

Acceptance. FULLY RESOLVES: over the public CLICS3 release, for each colexified concept pair report the count of distinct families and macroareas attesting it plus a family/area-stratified permutation significance; output a ranked list of control-surviving colexifications with effect sizes, and a comparison of the raw-weight ranking vs. the control-adjusted ranking (identifying confounded edges); report the network degree distribution and whether the strongest edges are confounded; ship a reproducible pipeline from public CLICS/CLDF data. PARTIAL: the family/area breakdown for the top-N CLICS colexifications, or a controlled test of a specific proposed universal colexification (e.g. SEE/KNOW, HAND/ARM, MOON/MONTH).