How similar are pitch contours derived from 'imaginary' student interactions to those derived from real interactions?

Abstract

This study evaluated the use of imagined interactions in speech research, by comparing speech addressed to imaginary speech partners with natural speech addressed to genuine interaction partners. The shape of pitch contours derived from target words of samples of imaginary infant (IDS), foreigner (FDS) and British adult (ADS) directed speech produced by ten female students was compared to that derived from an existing data set of natural IDS, FDS and ADS. For the pitch contour shape analysis we used a standard qualitative approach and a previously evaluated novel algorithmic method. We found no significant difference in pitch contour shape between IDS, ADS and FDS in the imaginary interactions. Unlike our previous findings for natural speech where IDS had been characterised by exaggerated contours compared to both adult conditions, all three speech types had a similar distribution of pitch contour shape. This contrast between the present findings and the interactions with genuine speech partners, suggests that speech obtained from imaginary interactions should be used with caution.