The pitfalls of negative evidence

Abstract

Beginning with publications in the early 1980s there have been attempts to use syntactic data to determine the highest-order subgroups of Austronesian. These efforts fall into two categories: those which claim that the voice affixes of Philippine-type languages originally had exclusively nominalizing functions, and those which claim that the affixes themselves were innovated after the separation of Rukai from the ancestor of all other Austronesian languages. Although these ideas lay dormant for some years, recently both have been revived in renewed efforts to show that the Austronesian family tree is not ‘rake-like’ in its highest nodes, but shows extensive embedding of subgroups that can be justified by successive layers of syntactic innovations. This paper questions the methodological soundness of both types of arguments on the grounds that they appeal to negative evidence, and logically any such appeal can do no better than reach an inference of indeterminate status rather than the positive conclusions that have been proposed.