Phonetics and phonology in FDG: Interfaces, mismatches, and the direction of processing

In this paper we discuss the interfaces between phonological and phonetic representations in Functional Discourse Grammar, and the possible mismatches that occur at those interfaces. Firstly, we discuss different definitions of phonological opacity in the literature, and provide examples with these definitions. We argue that mismatches between phonological and phonetic representations can result from competing pressures of articulatory ease and perceptual distinctivity. In order to model these influences and the resulting mismatches adequately, the model should not be organised strictly top-down: we argue that FDG should incorporate bottom-up influence from the phonetics on the phonology. We show that these influences are language-specific, which entails that bottom-up feedback must involve the Grammatical Component. With this modification of the model’s architecture, language users’ tendency to speak efficiently can be incorporated into the model, explaining a wide array of phenomena such as (synchronic) reduction, the cross-linguistic frequency of phonological alternations, and (diachronic) grammaticalization.