A cross-linguistic study of speech modulation spectra

Abstract

Languages show systematic variation in their sound patterns and grammars. Accordingly, they have been classified into typological categories such as stress-timed vs. syllable-timed on the basis of their rhythms, Head-Complement vs. Complement-Head on the basis of their basic word order, or tonal vs. non-tonal on the basis of the presence/absence of lexical tones. To date, it has remained incompletely understood how these linguistic properties are reflected in the acoustic characteristics of speech in different languages. In the present study, the amplitude-modulation (AM) and frequency-modulation (FM) spectra of 1862 utterances produced by 44 speakers in 12 languages were analyzed. Overall, the spectra were similar across languages. However, a perceptually based representation of the AM spectrum revealed significant differences between languages. The maximum value of this spectrum distinguished between HC non-tonal, CH non-tonal, and tonal languages, while the exact frequency of this maximum value differed between stress-timed and syllable-timed languages. Furthermore, when normalized, the f0-modulation spectra of tonal and non-tonal languages also differed. These findings reveal that broad linguistic categories are reflected in differences in temporal modulation features of different languages. This has important implications for theories of language processing and acquisition.