Introduction

This chapter will necessarily be rather different from most of those that precede, since it deals with languages of which no direct record survives – languages which are, by definition, prehistoric. Prehistoric languages can only be studied inferentially, and the only sound basis for our inferences is the well-known uniformitarian principle (UP). As applied in historical linguistics, the UP states the following:

Since the only alternative is unconstrained guesswork, all scientific historical linguists must take the UP seriously. It follows that we must not interpret what we find in the written historical record in any way that is inconsistent with the range of structures and changesin-progress that we can observe in languages currently spoken, nor must we posit for a prehistoric language any type of structure or change that is not actually attested somewhere among the known languages of theworld. This very general principle has remarkably specific consequences, especiallywith regard to phonological change (see §3 below), which constrain and guide our hypotheses with a precision often surprising to interested observers outside the field.