Published online by Cambridge University Press: 01 January 2026
We modeled the Greek H*, L+H*, and H*+L pitch accents using functional principal component analysis, followed by statistical modeling and curve reconstruction. The accents were distinguished by F0 height and shape. The data also exhibited cue trading between F0 and duration, as well as systematic context-driven variation and general variability, which led to category overlap comparable to that reported for vowel contrasts. These findings indicate that intonation categories are more similar to segmental categories than previously thought, supporting the view that the study of intonation phonetics and phonology should follow the same principles as the study of segments.
We thank our participants, who generously donated their time to this project, Georg Lohfink and Yang Li for preliminary data processing and analysis, Katherine Marcoux for data analysis and the recordings of our English examples, and Mary Baltazani for sharing her F0 error detection script with us. We also thank two anonymous referees, and Andries Coetzee, Shelome Gooden, and Anne-Michelle Tessier for their helpful comments and suggestions. The financial support of the European Science Foundation through grant #ERC-ADG-835263 to Amalia Arvaniti, and intramural support from the University of Kent to Amalia Arvaniti in the early stages of this work, is hereby gratefully acknowledged.