Quantitative modelling and syntactic microvariation

Abstract

This paper argues that parametric theories should account for statistically significant associations between syntactic variables that are logically and historically independent. Associations, which do not necessarily imply causation, suggest that syntactic variation is limited by abstract constraints (Parameters), which are not categorical rules. Counterexamples, which are always found in microvariation contexts, do not necessarily lead to the refutation of solid associations and, consequently, to the refutation of parametric models. It is argued that parametric models should be sufficiently plastic to adapt to unexpected patterns that are learnt from experience.