Dynamic model of speech

This article explores the relationship between intonational patterns and its relationship with speech rhythm and discourse, according to the dynamic systems research program. The study of these relationships were based on Barbosa’s (2006) Dynamic Model of Speech Rhythm; on Dato intonational annotation system proposed by Lucente (2008); and on the Computational Model of the Structure of Discourse, proposed by Grosz & Sidner (1986). The Dynamic Model of Rhythm suggests that the speech rhythm is the result of the action of two oscillators – accentual and syllabic - which receive as input linguistic and gestural information, and give the output as gestural duration. This article hypothesis is that in addition to these oscillators, a glottal oscillator can act controlling the intonation patterns of speech. These patterns, or intonational cycles, which organize the BP intonation, emerge when related to the spontaneous discourse segmentation. For each discourse segment classified as spontaneous according to a criteria proposed in this article, the speech is segmented into the DaTo system in linguistically structured units, which contains the purposes of communication and attention. Each of these segments is aligned to the speech intonation pattern delimitated by a rising contour (LH or> HL) at the beginning and by a falling contour (LHL), or a boundary level (L), at the end. The speech rhythm is also aligned to the pattern formed between intonation and discourse. By the inclusion of a new layer for the stress groups segmentation into DaTo system was possible to observe that the alignment between the stress groups segmentation and the intonational annotation coincide with discourse segments boundaries. The alignment between intonation, rhythm and discourse, having the stress groups as attractors, allowed us to propose the insertion of a glottal oscillator into the Dynamic Model of Rhythm.