Does intonation have primitive units?

Abstract

This study investigated whether listeners can interpret intonation based on partial rather than global fundamental frequency (F0) or pitch information. Matched Statements and echo questions were recorded digitally at 22 kHz sampling rate using a Computerized Speech Lab (CSL). Each production was stored on the CSL and edited by slicing the wave form into two parts that corresponded to terminal (last 150–200 ms span of the F0 contour) and preterminal (section of contour preceding the terminal region) F0 contour regions. A total of 92 stimuli was thus generated and recorded on a DAT tape. The tape was played to 96 young (20–28 years) and old (42–79 years) adults to determine if each stimulus was part of a statement or a question. For preterminal (PTL) stimuli, both groups successfully identified statements 91% and 88% of the time, respectively, compared to 72% each for questions. The respective scores for the terminal (TL) stimuli were 94% and 91% (statements) versus 91% and 87% (questions). Across groups, statements were easier to identify with both TL (p=0.002) and PTL (p=0.001) cues compared to questions. These findings suggest that intonation decoding might involve units smaller than global F0 contour. [Work supported by NIH (NIDCD) grant 5R03DC04955-02.]