This book explores properties of mental grammars using evidence from diachronic change and synchronic alternations and proposes a lexicon- and usage-based model of phonology and morphology. The central role in this model is afforded to the lexicon, which consists of stored surface-based representations. Morphophonological generalizations are extracted from phonetically-detailed lexical representations and represented in terms of schemas, whose strength is calculated on the basis of type frequency (i.e. the number of instances that observe a given generalization). Another factor that emerges from usage and is taken into account is token frequency (i.e. the number of instances of use of a particular item) and its impact on the strength of mental representations.
A careful analysis of morphophonological patterns and their conditioning factors reveals that schemas have access to morphological, morphosyntactic and semantic information. Phonetic and psycholinguistic factors also affect morphological patterning. It is argued that redundancy-free approaches are inadequate in the light of irregularities that are inherent to linguistic systems. Grammars are uneconomical and do not reflect the requirements of an optimal coding system. The overarching goal of this proposal is to integrate generative and lexicon/usage-based models of grammar.