How do voiceless fricatives contribute to intended intonation? a comparison of whispered speech, semi-whispered and normal speech

The overall goal of the paper is to contribute to a better understanding of segment-prosody interaction in different speech modes. In particular, the paper shows that depending on intonation patterns, utterance-final voiceless fricatives in Polish show different properties. In polar questions with rising intonation, the frequency of the highest spectral peak is higher than in statements produced with falling intonation. Voiceless fricatives in questions display higher Centre of Gravity accompanied by higher standard deviation. Furthermore, the fricatives are produced with higher mean intensities in questions than statements. Differences in fricatives are more pronounced when the sentences are whispered. For instance, in contrast to normal speech where skewness and kurtosis do not show differences, in whispered speech both spectral moments display lower values in statements than in questions. In addition, spectral slopes (m1, m2) are steeper in questions than in statements. In summary, the paper shows that voiceless fricatives are prone to intonation differences and follow its patterns by means of varying spectral cues inherent to these sounds.