Detecting contact in language trees: a Bayesian phylogenetic model with horizontal transfer

Abstract Phylogenetic trees are a central tool for studying language evolution and have wide implications understanding cultural as whole. For example, they been the basis of studies on musical instruments, religious beliefs political complexity. Bayesian phylogenetic methods transparent regarding data assumptions underlying inference. One these assumptions—that languages change independently—is incompatible with reality evolution, particularly contact. When speakers interact, frequently borrow linguistic traits from each other. ignore this issue, which can lead to errors in reconstruction. More importantly, neglect rich history A principled way integrating contact is sorely missing. We present , model horizontal transfer evolution. The efficiently infers tree family events between its clades. implementation available package phylogenetics software BEAST 2. apply simulation study case subset well-documented Indo-European languages. demonstrates that correctly reconstructs simulated family, including events. Moreover, it shows ignoring systematic estimated height, rate topology, be avoided . confirms known finds loanwords, demonstrating practical potential. has higher statistical fit than conventional reconstruction, reconstructed height significantly closer well-attested estimates. Our method closes long-standing gap theoretical empirical models especially relevant less documented families, where our knowledge past contacts borrowings limited. Since phylogenies become backbone many addition integral piece puzzle crucial endeavour understand human culture.