Mar 19, 2009

2009/03/19

Pierrehumbert, J. (1979). The perception of fundamental frequency declination. Journal of Acoustical Society of America, 66(2), 363–369.


Presentation: Shelly
Summary: Sally

Previous studies on declination mainly focused on the physiological mechanism and more natural synthesized speech. In this study, effects of pitch range, amplitude, as well as utterance length were examined. Materials used were different manipulations from the original sentence “The baker made bagels,” with all syllables being replaced by ma’s and the basic prosodic pattern (2 pitch peaks) being preserved.

In the pitch range experiment, two pitch ranges were given to the subjects to judge for correction of declination. Results showed that greater correction was made for declination with a wider range. Two experiments were conducted for amplitude. In the first, the subjects were asked to judge which of the two stressed syllable was higher in pitch. Results showed that increase of the amplitude on a peak would increase the impression that its pitch was higher. A second experiment was then carried out in order to check whether subjects could tease the effect of amplitude and pitch apart. They were asked to judge loudness of the two peaks first, then their relative pitch. Results showed that the amplitude effect was only half as strong as that obtained in the first experiment, indicating that the subjects actually relied on the relative “prominence” (a combination of amplitude and pitch) when making the decision.

The last factor being examined was utterance length. Again, two experiments were conducted. In the first, utterance length was varied by inserting one (set-1: short), two (set-2: medium), or three (set- 3: long) unstressed syllables between the two stressed ones. Results showed that in general, the slope of expected declination was less steep for longer stimuli. However, to rule out the possibility that declination might still vary with the number of “stressed” syllables, a second experiment was carried out. Stimuli were created by having three stressed syllables in the utterance. Two relative pitch heights were included to avoid the subjects comparing the second peak with this newly-added stressed syllable. It was confirmed that increasing stressed syllables did not increase the expected declination.

No comments: