Proceedings of the Xth Conference, Helsinki 2008
Edited By Martti Vainio, Reijo Aulanko and Olli Aaltonen
25. Correspondences between KIM-based symbolic prosodic labels and parameters of the Fujisaki model 261
CORRESPONDENCES BETWEEN KIM-BASED SYMBOLIC PROSODIC LABELS AND PARAMETERS OF THE FUJISAKI MODEL Hartmut R. Pﬁtzinger Hansjörg Mixdorff Benno Peters 1 Introduction The current study examines the relationship between prosodic labels assigned in the Kiel Corpus of Spontaneous Speech IV and parameters of the Fujisaki model of the production process of f0 (Fujisaki and Hirose 1984). By applying a quantitative model to f0 contours extracted from spontaneous speech of four speakers we make f0 peaks accessible to empirical analysis of the temporal alignment with segmental landmarks, e.g. the vowel onset of the accented syllable, and of the relationship between f0 interval and accent level class. Tonal alignment is a research topic at least since Bruce 1977 who investigated the two distinctive prosodic patterns connected with primary stressed syllables in Swedish. He found temporal stability of the f0 contours. More speciﬁcally, “an early timing of the fall is interpreted as accent I and a late timing of the fall as accent II with a sharp shift of identiﬁcation” (Bruce 1977, p. 147). Recently, the question was investigated whether alignment differences across languages arise from a continuum of phonetic alignment realizations which fall within a single phonological category (Atterer and Ladd 2004) or are realizations of several phonological categories (Niebuhr and Ambrazaitis 2006). Mixdorff and Fujisaki 2000 compared German ToBI labels with Fujisaki param- eters. They found that tone labels were strongly correlated with accent commands, and the type of label (typically H*L and L*H) was clearly...
You are not authenticated to view the full text of this chapter or article.
This site requires a subscription or purchase to access the full text of books or journals.
Do you have any questions? Contact us.Or login to access all content.