Edward Sapir notably used evidence of contact and diffusion as a negative tool for genetic reconstruction, treating it as a subject in its own right only at the end of his career (e.g., for the influence of Tibetan on Tocharian).
[2] William Labov in 2007 reconciled the tree and wave models in a general framework based on differences between children and adults in their language learning ability.
Adults do not preserve structural features with sufficient regularity to establish a norm in their community, but children do.
[3] Diffusion of areal features for the most part hinges on low-level phonetic shifts, whereas tree-model transmission includes in addition structural factors such as "grammatical conditioning, word boundaries, and the systemic relations that drive chain shifting".
Some examples are the Balkan sprachbund, the Mainland Southeast Asia linguistic area, and the languages of the Indian subcontinent.