Artificial Lineage vs. Physical Diffusion Rethinking Language Classification Beyond the Family Tree Model

This paper proposes a new conceptual framework for language classification by distinguishing between Artificial Lineage Languages and Physical Diffusion Languages. The former refers to languages that experienced large-scale political or institutional unification—such as Classical Latin or Classical Chinese—before fragmenting into descendant languages, making tree-based models effective. The latter refers to language communities that never underwent such unification and instead spread through natural geographic and social diffusion, forming dialect continua and contact networks rather than genealogical branches. By focusing on Tungusic and Ainu language ecologies in Northeast Asia, the paper argues that tree models fail in historically diffuse environments and should be replaced by wave, network, and field-based approaches. This distinction explains why some language families resist clean genealogical classification and provides a theoretical foundation for modeling language change as a spatial–temporal diffusion process rather than as a linear descent.