Christopher A. Moore

Quantitative description and differentiation of fundamental frequency contours

Fundamental frequency (f0) contours derived from the speech of 35 mothers to their 4-month-old infants were quantified for two experimental conditions, one in which the mother was instructed to seek her infants attention and a second in which the mother was instructed to express approval of her infants action. In addition to conventional descriptions (e.g. mean and standard deviation of f0, and utterance duration) the contours were subjected to modelling using 16 equations (1 linear and 15 non-linear) selected to reflect customary qualitative descriptions of f0 contours (e.g. Gaussian, rising, falling). Curve-fitting results confirmed that these infant-directed utterances were fit extremely well by at least one function. The average maximum R2 value obtained across all 16 equations was 0·83. Furthermore, discriminant analysis demonstrated that these two utterance types could be differentiated with 76% accuracy by these curve-fit results alone. Discrimination improved to 92% accuracy, however, when additional, more global descriptors were included in the discrimination function (e.g. utterance duration, mean fundamental). These results suggest that infants may be responsive to specific prosodic stimuli, which may involve distinct voice dynamics or more general speech signal characteristics, such as overall pitch, pitch variability, or utterance duration.

Wait-time in college classes taken by education majors

Professors teaching education classes differed little from professors teaching noneducation classes in terms of the questions they asked and how long they paused after questions and students responses. Fewer professor questions went unanswered in the education courses. Professors ask on average about 25 questions each class hour, the majority of which are higher-level. They pause about 2.25 seconds after questions and .45 seconds after student responses. These pauses suggest many students are effectively shut out of responding and are not provided the opportunity to elaborate their answers, even those to complex, divergent questions.

Tests of the precedence effect in sound localization reveal abnormalities in multiple sclerosis

The precedence effect in sound localization involves presenting identical sounds (e.g., clicks) from pairs of matched speakers situated on opposite sides of a subjects head, with the clicks from one speaker preceding those from the other by a short interval. With appropriate delays, normal subjects perceive a fused image which originates from the side of the leading speaker. This test was administered to 24 patients with multiple sclerosis (MS). Separate tests involving speaker delays ranging from 0 msec (simultaneous presentation) to 8 msec were presented. At 0 msec delay, normal subjects perceived the fused image to be located halfway between the two speakers; at progressively longer delays, the image was perceived closer to the leading speaker. In contrast to normal subjects, a large proportion of the MS subjects exhibited difficulties with the task. The discrimination deficit was limited to delays below 1 msec, suggesting a problem involving an increased threshold for lateralizing the fused image away from midline toward the side of the leading speaker. The neural instability produced by demyelination in MS patients might account for this pattern of results.