[Part I. Challenges to accepted views of convergence and divergence in language contact situations, Divergence, convergence, contact]
The genealogical paradigm, classically formulated by Meillet, still dominates language classification. Despite of its weaknesses, the tree model is at the base of all global or particular language lists. Distinguishing common heritage from subsequent borrowing is a matter of hot debates in all parts of the world. This contribution challenges the universal validity of this paradigm. Genealogical trees are based on divergent evolutions, leading to the formation of new species. But this biological metaphor has its limits when applied to language. The formation of a new language is only a “pseudo-speciation” (Ericson), communication remains possible even after the split, and convergence may counterbalance the effects of divergence at any moment. The evolutionary model of Indo-European does work in some cases, but not in all.