A model for predicting the prosody of spontaneous speech (PPSS model)

Abstract

Listen

Translate article

The prosodic structure of speech is the result of complex interactions within and between several different levels of organization. The intonative hierarchy, which is essentially manifested by the nature of the prosodic markers, is the product of complex interactions and constraints within and across organizational levels. Presented here is a model for predicting and interpreting the prosodic organization of spontaneous speech utterances. This model is a hierarchical system composed of six modules: (1) sematic-pragmatic, (2) syntactic, (3) phonotactic, (4) accentual, (5) semantic adjustment, and (6) rhythmic. For a given utterance, the system determines (i) the levels of the boundaries and prosodic markers on the basis of semantic information, and the syntactic structure as defined by the X-bar theory, and (ii) the accentual and rhythmic structures based on phonotactic constraints. The phonetic step, which should transform the abstract labelling into acoustic values, is not presented here. This model can and should be further developed. Future enhancements will pertain to (a) the nature of the rules, (b) different aspects of conversation, and (c) theoretical considerations. Concerning the latter point, current fruitful developments in X-bar theory are likely to lead to positive modifications in the prosodic model, which should enable it to account for certain unexplained phenomena. However, even in its current state, the model produces highly convincing results, since it predicts the number and hierarchy of intonative and stress units in an utterance with a high accuracy rate.