Morphological Universals and Diachrony

Although linguistics is plausibly taken to be “the science of language,” the actual object of inquiry in the field has changed considerably over time. Prior to the influence of deSaussure in the first part of the twentieth century, linguists concerned themselves primarily with the ways in which languages have developed historically. For the next several decades, they devoted their attention to the external facts of sounds, words and sets of utterances. With the advent of the cognitive (or “Chomskyan”) revolution around 1960, however, they came increasingly to see themselves as studying the human language faculty: speakers’ knowledge of language and the cognitive capacity that makes this possible (Anderson & Lightfoot 2002), Universal Grammar. This is what our theories attempt to represent nowadays. Unlike the documented facts of language history or the measurable properties of sounds and utterances, such a cognitive faculty is not directly observable, so the question naturally arises of how we might study it empirically. Two important modes of argument have emerged that are generally taken to aid in this enterprise. First, if we can show that speakers know something about their language for which relevant evidence is not plausibly present in the input on the basis of which they learned the language, we assume that this knowledge must be a consequence of the structure of the ‘language organ’. This is the argument from “the poverty of the stimulus,” and (despite the skepticism of some: e.g. Pullum & Scholz 2002) it has proven to have wide applicability, especially with respect to speakers’ knowledge of syntax. A second line of argument is to assume that when we find that something is true of all (or at least nearly all) of the languages we can observe, it must be true of Language more generally, and thus a property of the human language faculty. The assumption that valid generalizations about language typology must be reflected in