An acoustic functional grammar of English intonation

Abstract

There has been a resurgence in recent years of linguistic models that directly embed exact physical and acoustic data right into the formalism. This work attempts to show that many aspects of speech perception and production can only be understood when the acoustic input is fully considered. In the case of speech perception, if there are abstractions, they are abstractions over acoustic input, not based on a priori phonological categories. The current paper extends this work into English intonation. The functional phonology of Paul Boersma, which incorporates exact frequency values into an optimality theoretic perception grammar, is extended to explain several basic facts about English intonation, including the derivation of categorical high and low tones from the acoustic signal, the relation of tones to hierarchically adjacent tones, and the affectual aspects of classically paralinguistic intonation. The primary advantage of this approach is that it allows a single model to handle both linguistic and paralinguistic intonation structure, a merger which is only possible due to the use of physical acoustic data in the model.