講題：Why Can’t Bayesianism Make Sense of An Evidential Methodology in Historical Linguistics?
時間：2021.3.24 (Wed.) 15:10-17:00
Abstract: The Bayesian approach is widely adopted in scientific practice and is helpful for philosophers to spell out the methodology of scientific reasoning. This approach fares well in well-specified settings, where prior probabilities are empirically defensible or will be overridden hence the insignificance of their influence on the posterior probabilities. However, in poorly specified settings such as historical linguistics, Bayesianism hardly fits the methodology. Using the reconstruction of proto-sounds as a case study, I argue that prior probabilities and catch-all likelihoods are generally so scarce in such settings that the Bayesian approach fails to make sense of the rationale behind the comparative method.