Edited by Ritsuko Kikusawa and Lawrence A. Reid
[Current Issues in Linguistic Theory 326] 2013
► pp. 153–172
It is an ordinary practice in comparative linguistics that idiosyncratic common features observed between historically related languages are ascribed to their parent language. The possibility of accidental similarity is usually rejected in situations where more than one language shows a unique morphological correspondence which is impeccable in terms of form and meaning. However, an apparently unmistakable correspondence of this kind sometimes turns out to be a mirage resulting from parallel and independent developments in the internal histories of the relevant languages. Such a case is shown by languages which have numerous documents written over many centuries. In this paper I will attempt to demonstrate that some features traditionally posited for Proto-Indo-European are in fact mirages. Needless to say, the comparative method is a powerful tool for reconstructing proto-languages, and there is a constant temptation when practicing the comparative method to attribute too much to the common ancestor. It is therefore important to recognize its limitations.