Comparative Reconstruction Probabilistically: The Role of Inventory and Phonotactics

Abstract

I introduce a novel quantitative methodology for evaluating manual comparative reconstructions. This method is incumbent on the existence of a manual comparative reconstruction and, unlike previous quantitative methods, cannot give a result contradictory to the reconstruction. The primary goal for this framework is to reconcile traditional and quantitative methodologies and act as an objective and accessible platform for comparative reconstruction, thereby extending the scope of historical linguistics further into the past. A few theoretical corollaries of the framework are also presented. It is shown that the likelihood that a reconstruction is spurious is related to some of the phonological properties of the descendent language. This likelihood is inversely correlated with mean word-length and segmental inventory size. Additionally, most active phonological processes and cooccurrence restrictions in the language – such as phonotactic constraints, prosodic effects, segment harmony, and neutralization – all serve to increase the likelihood that a reconstruction to that language is spurious.