Measures of phonation types in a natural language

Abstract

Using a Kay digital spectrograph, three types of analysis (spectrograms, power spectra, and waveforms) were used to investigate different phonation types in a single natural language (Jalapa Mazatec). In addition to normal voice, this language uses creaky voice and breathy voice to distinguish words. We recorded five speakers producing contrastive vowels that were identical in quality and pitch. The spectrograms showed that in comparison with normal vowels the creaky vowels had glottal pulses at irregular intervals and a greater amplitude in the higher frequencies. The breathy vowels are more clearly distinguished in the onset, and they often lacked discernible pulses. We used power spectra displays to quantify phonation type differences. In creaky vowels the amplitude of the first harmonic in relation to the amplitude of the first formant is less than in normal voice; and in breathy vowels the first formant had less intensity in comparison with that of the first harmonic. Waveforms displayed the irregularities in the glottal pulses and the high degree of first formant damping in creaky vowels. [Work supported by USPHS grant NS 18163‐02.]