Soliloquy Learning's Reading Assistant is a speech-recognition based software system that is designed to present text and graphics on the computer so as to engage, assist, and assess students' oral reading. In Phase 1 of this project samples of oral reading fluency were collected from 196 students in grades 2 through 6 and across a range of passage-difficulty levels. Analyses of these data focused on interactions between passage level and reading ability. The data and analyses were also used to tune and test the performance of the Reading Assistant's speech recognition layer. The software's ability to discern competent from unacceptable readings exceeded 95% for both English language learners and native speakers of English. In Phase 2 of the project, Soliloquy Learning plans to evaluate whether regular, classroom use of the Reading Assistant effectively promotes children's reading growth. Treatment students will be asked to read with the software for one hour each week. To examine the impact of the software across demographic settings, the study will involve Grade 2-4 students in inner-city and in median-income, suburban schools. To examine the extent to which benefits of the software extend to older children with longer histories of reading struggle, the study will also include a treatment and control sample of middle-school students. To investigate how the impact of the software may accrue, plateau or change in nature with longer-term use, treatment for both elementary and middle-school cohorts will extend across two school years. Dependent measures include word recognition, fluency, vocabulary, and full-scale test scores. In addition, audio files and data logs created by the Reading Assistant during each reading session will be used to monitor usage and progress more closely. In complement to the intervention study, Soliloquy has proposed several targeted investigations: of reading characteristics, speech timing, and prosodies as they relate to reading fluency, of the nature of textual difficulties that trigger disfluencies, and of the relation of comprehension and fluency measures in assessing overall reading proficiency. ? ?
