Key points are not available for this paper at this time.
In recent literature on persons with learning disabilities (LD), speech recognition has been addressed primarily as an assistive technology to compensate for writing difficulties. However, prior research by the authors has shown that speech recognition may have a remedial effect on reading and spelling. The present study was conducted to compare the effects of two types of speech recognition systems, discrete speech and continuous speech, in terms of their relative remedial effects. Thirty-nine students with LD (aged 9-18) had participated in a previous study. Nineteen had used discrete speech systems for 50 minutes a week for sixteen weeks to write on self-selected topics, while twenty students in a control group had received general computer instruction. In the following semester, an additional experimental condition was added. Thirteen participants used continuous speech recognition systems 50 minutes a week for sixteen weeks to write on self-selected topics. Results were then compared across all three groups. Both discrete and continuous speech groups showed significant improvement in word recognition (p<.002, p<.018) and reading comprehension (p<.043, p<.041) over the control group and the discrete condition also showed significant improvement in spelling (p<.015). Pre- and post-tests on five reading-related cognitive processing measures (phonological, orthographic and semantic processing, metcognitive reading strategies, and working memory) indicated that only phonological processing improved significantly, and only for the discrete speech group when compared to controls (p<.04). Results were related to specific differences between continuous and discrete speech systems in terms of the task requirements for generating and correcting text using the two systems.
Higgins et al. (Wed,) studied this question.