Universals constrain change; change results in typological generalizations Paul Kiparsky Stanford University 1 The relation between synchrony and diachrony 1.1 Structure explains change If language change is constrained by grammatical structure, then synchronic assumptions have di- achronic consequences. Theories of grammar can then in principle contribute to explaining prop- erties of change, or conversely be falsified by historical evidence. This has been the main stimulus for incorporating historical linguistics into generative theorizing. A widely shared assumption is that certain mutations occur in the transmission of language. Specifically, they occur when aspects of grammars based on incomplete data, or outputs of such grammars, can be retained from earlier stages of acquisition and become incorporated into the final system. This notion of “imperfect learning” has provided the basis for one approach to analogical change, and, coupled with the theory of Lexical Phonology, provides a solution to the problematic type of phonological change known as lexical diffusion (Kiparsky 1995). It is also commonly assumed in investigations of syntactic change. The theory of acquisition thereby becomes a crucial link between synchronic and diachronic linguistics. The specific implementation of this approach will depend on the model of grammatical de- scription that is adopted. Syntactic change, for example, has been treated as parameter-resetting (Lightfoot 1991), as grammar replacement (Kroch 1989), and as constraint reranking (OT, recently especially in its stochastic variety, J¨ ager & Rosenbach 2003, Clark 2004). Each comes with dif- ferent commitments about the causes and mechanisms of change and about how change is related to synchronic variation. Specific theories of syntax make further predictions about co-variation between different aspects of grammar, notably between morphology and syntax. For example, on some versions of syntax, rich inflectional morphology entails a highly ramified structure of functional categories to which categories move to check their features, predicting that loss of verb agreement entails loss of V-to-I movement (e.g. Vikner 1995). In a different framework, I have ar- gued that structural position and inflectional morphology are alternative argument licensers, from which I derive, among other consequences, the Sapir/Jespersen generalization that loss of inflec- tional morphology entails fixed order of direct nominal arguments (Kiparsky 1997). The leading idea behind this work, that properties of language change might be explained by the way language is acquired and structured in the mind, is of course by no means original to generative grammar. The neogrammarians, for example, had recognized the pervasive role of analogy as a regularizing force in change as a manifestation of the mechanism that underlies the 1