Automatic assessment of expressive oral reading
Speech Communication
| Hi-index | 0.00 |
This study assessed the effectiveness of software designed to facilitate expressive oral reading through text manipulations that convey prosody. The software presented stories in standard (S) and manipulated formats corresponding to variations in fundamental frequency (F), intensity (I), duration (D), and combined cues (C) indicating modulation of pitch, loudness and length, respectively. Ten early readers (mean age=7.6years) attended three sessions. During the first session, children read two stories in standard format to establish a baseline. The second session provided training and practice in the manipulated formats. In the third, post-training session, sections of each story were read in each condition (S, F, I, D, C in random order). Recordings were acoustically examined for changes in word duration, peak intensity and peak F0 from baseline to post-training. When provided with pitch cues (F), children increased utterance-wide peak F0 range (mean=34.5Hz) and absolute peak F0 for accented words. Pitch cues were more effective in isolation (F) than in combination (C). Although Condition I elicited increased intensity of salient words, Conditions S and D had minimal impact on prosodic variation. Findings suggest that textual manipulations conveying prosody can be readily learned by children to improve reading expressivity.