Use of text-to-speech software to improve reading

Available online at www.sciencedirect.com

Procedia Computer Science 14 (2012) 359 – 362

International Conference on Software Development for Enhancing Accessibility and Fighting Info-exclusion (DSAI 2012)

Use of text-to-speech software to improve reading skills of high school struggling readers Robert A. Stoddena*, Kelly D. Robertsa, Kiriko Takahashia, Hye Jin Parka, and Norma Jean Stoddena* a

University of Hawaii at Manoa, Center on Disability Studies 1776 University Avenue, UA4-6 Honolulu, HI 96822, USA

Abstract The purpose of this paper is to present research findings on the effectiveness of a TTS software intervention in two pilot studies with approximately 104 high school students (grades 9-12) with a disability and at-risk for referral to special education services because of reading difficulties. The conceptual framework underlying the research is that the use of TTS software with content y academic performance and aspirations. The are not using the software. Improved reading is also expected to improve su results indicated that study participants had significantly improved reading skills in the two pilot studies. The researchers abulary learning through the use of the TTS software.

Keywords: Text-to-speech (TTS); Learning Disabilities; Struggling Readers; High School

1. Introduction The complex task of reading is one of the most commonly cited reasons for high dropout rates in US schools [1] [2]. Many high school students struggle with reading, especially reading comprehension, because of their slow rate and inaccurate decoding. In addition, many students lack sufficient content area vocabulary. This vocabulary gap widens over years in school with 12th grade, high performing students knowing about four times as many words as low-performing students in comparison to high performing first graders who know about twice as many words as low-performing students [3]. Students who do not have the reading skills or access to improve their reading skills by high school may be more likely to dropout because they cannot keep up with coursework. Text-to-speech (TTS) software is one example of AT that has become a more common tool for struggling readers in high schools and colleges, and has been widely accepted as a form of accommodation for students with

*

Corresponding author. Tel.: +1.808.956.9199; fax: +1.808.956.3162 E-mail address: [email protected]

1877-0509 © 2012 2012 The Authors. Published by Elsevier B.V. Selection and/or peer-review under responsibility of the Scientific Programme Committee of the 4th International Conference on Software Development for Enhancing Accessibility and Fighting Info-exclusion (DSAI 2012) doi:10.1016/j.procs.2012.10.041

360

Robert A. Stodden et al. / Procedia Computer Science 14 (2012) 359 – 362

disabilities [4] [5]. However, research [7]. The purpose of the research conducted in the two pilot studies and presented in this paper was to determine the efficacy and potential of TTS software (i.e., Kurzweil 3000) [8] as a reading intervention for high school struggling readers. 2. Research Questions In the first pilot study, the following two research questions (RQ) were investigated: 1. standardized vocabulary test (unaided)? 2. How did the use of TTS software over a period standardized reading comprehension test (unaided)? 3. Building upon the findings of the first pilot study, the following additional question was investigated in the second pilot study: How did the use of TTS software over the period of one semester affect the unaided reading rate of the subjects? 3. Method The use of Kurzweil 3000 TTS software as a reading intervention was conducted in the two pilot studies. Both studies utilized mixed methods [9] [10] to address the research questions. The research design of the two pilot studies was a repeated measures design, where the baseline and post-scores of each subject were compared to themselves. 3.1. Intervention The TTS software used for this study was the Kurzweil 3000 which was designed as a compensatory tool for individuals with disabilities, including learning disabilities (LD), attention deficit hyperactivity disorder (ADHD), reading difficulties, and some physical disabilities, such as quadriplegia where turning the pages of a book is difficult. TTS software provides a synchronized visual and auditory presentation of text. It plays back text as spoken words from almost any good quality printed document (e.g., books, articles, and magazines) that has been converted into an electronic text format, or from electronic documents. At each participating school, the Kurzweil 3000 was installed on the network servers. All students who returned signed student and parental consents indicating agreement to participate had access to the software on classroom computers and in computer labs. 3.2. Procedures of the First Pilot Study Thirty-five Hawai`i public high school students, who had been diagnosed with learning disabilities and were reading between a 1.0 and 6.0 GLE as measured by select subtests of the Woodcock-Johnson III (WJ III) Test of Achievement (1989)[11], participated in the first pilot study. The subjects were trained on the use of the software and a fidelity checklist was used to determine that each could use the required features of the TTS software with 95% accuracy. The TTS software was used by the subjects to read class reading material and assignments, either with supervision from the researchers or independently, for one semester. The WJ III was administered to measure any change in unaided reading comprehension and vocabulary three times to the subjects: before the intervention, at the end of the first quarter, and at the end of the second quarter/semester. To avoid a practice effect, alternate forms of the WJIII were administered. Complete data were obtained for 17 of the students. Findings from the first pilot study were used to make revisions for the second pilot study. The revisions included development of: (1) a student survey to measure use and perception of TTS software; and (2) a TTS training manual


361

for high school students and teachers. The Nelson-Denny Reading Test replaced the WJIII as the measure of reading. This revision allowed for a group administration of the test and inclusion of reading rate as a measure. The change from an oral-verbal test to a pencil-paper based test was deemed more closely matched to typical high school reading assessments. The Nelson-Denny Reading Test was administered twice: pre- and post-intervention. 3.3. Procedure of the Second Pilot Study Sixtywere in grades 9 thru 12 and reading between a 1.0 and 6.0 GLE as measured by the Nelson-Denny Reading Test participated in the second pilot study. All subjects were trained on the use of the software and a fidelity checklist was utilized to ensure subjects could use the required features of the TTS software with 95% accuracy. The intervention consisted of students using the software for a minimum of 40 minutes per week for one semester, either with supervision from the researchers or independently, to complete class reading and assignments. The Nelson-Denny Reading Test was administered before and after the intervention. Alternate forms of the test were administered to avoid a practice effect. A student survey was also administered before and after the intervention. The student survey was developed after the first pilot study. Complete data were collected from ten of the students. 3.4. Data Analysis Quantitative data was analyzed using a general linear modeling for repeated measures using SPSS. To address the unbalanced design, the Type III Sum of Squares within SPSS was used. Qualitative data was analyzed using the constant comparative method of analysis [12]. Codes (i.e., free nodes and hierarchical nodes) were developed from the transcriptions, and the frequency of the codes and the relationship between codes and other variables were found. 4. Results

4.1. Findings from the First Pilot Study For research question (RQ) 1, the unaided vocabulary (vocabulary tested with WJ III when not using the software) of the students increased significantly over time at .10 level (F(2, 32)= 2.524, p=.096), and the partial eta squared value was .136, a medium effect size. For RQ 2, the unaided reading comprehension of the students did not significantly change over time at .10 level (F(2, 32)=.508, p=.607). Additionally, using a cluster analysis, the group of students with increased unaided vocabulary was found to have used the text-to-speech (TTS) software for approximately 40 (39.71) minutes per week on average in the first pilot study. Findings from the first pilot study included: (1) the use of TTS software in high school classrooms is feasible; (2) the group of subjects whose vocabulary scores increased at the end of the intervention period were those who used the TTS software for at least 40 minutes per week; (3) administering WJ III to individual students is inefficient; and (4) measuring student reading three times per semester is not an efficient method of capturing change over one semester. 4.2. Findings from the Second Pilot Study Significant improvements in unaided vocabulary (GLE score), unaided reading comprehension (GLE score), and unaided reading rates (percentile) of the students over time, at the .05 level, compared to their own baseline scores (t(9)=6.880, p=.000; t(9)=3.481, p=.007; t(9)=3.108, p=.013, respectively) were findings in the second pilot study.

362


The total reading scores significantly increased by 1.99 GLE on average at .05 level (t(10)=2.308, p=.046) at the end of the intervention semester, with a large effect size (d=.84). 5. Discussion The results indicated that study participants had significantly improved reading skills in the two pilot studies. The researchers attribute th the use of the TTS software. The software provided a steady pace of auditory and visual input of text for the struggling readers whose typical reading is slow and halting. The poor reading strategies used by participants were augmented and compensated by setting the reading speed of the software to a faster rate than what the participants could read (unaided) on their own. As participants became more familiar with the use of the software, they were encouraged to increase the speed. All participants set the software reading speed at 120 words per minute or higher. Moreover, when using the TTS software, the participants heard accurate and correct pronunciations of most substitution errors and decoding errors. According to literatures on reading, accurate and fast decoding of words will leave more room in working memory for constructing meaning as the reader reads a text [13] [14]. Thus, it is feasible to state that reading with the TTS software assisted students to be less fatigued and have more memory space to think about the actual content of what they were reading. 6. Conclusions In summary, participants who used the TTS software had improved reading rates, vocabulary, and comprehension. These gains are believed to be related to the increased exposure to text and incidental vocabulary learning that purportedly takes place when using TTS software. While there were a number of limitations to the study the researcher are optimistic about the potential. This includes the belief that through continual use of TTS software ongoing gains in reading vocabulary, comprehension, and fluency rates will continue to be made. References [1] M. L. Kamil, Adolescents and literacy: Reading for the 21st Century. Washington, DC: Alliance for Excellent Education, (2003). [2] C. E. Snow and G. Biancarosa, Adolescent literacy and the achievement gap:What do we know and where do we go from here? New York: Carnegie Corporation of New York, (2003). [3] B. Hart and T. Risley, Meaningful differences in the everyday experience of young American children. Baltimore: Paul H. Brookes Publishing, (1995). [4] E. Engstrom, Reading, writing, and assistive technology: An integrated developmental curriculum for college students. Journal of Adolescent and Adult Literacy, 49(1), 30-39 (2005). [5] H. Silver-Pacuilla and S. Fleischman, Technology to help struggling students. Educational Leadership, 63(5), (2006) 84-85. [6] A. Moorman, R.T. Boon, Y. Keller-Bell, C. Stagliano, and T. Jeffs, Effects of text-to-speech software on the reading rate and comprehension skills of high school students with specific learning disabilities. Learning Disabilities: A Multidisciplinary Journal, 16(1), (2010) 41-49. [7] D. Deshler, A Closer Look: Closing the Performance Gap. Stratenotes 13(4). University of Kansas Center for Research on Learning. Retrieved from http://www.adlit.org/article/19373 (2005). [8] undations workshop. Bedford, MA, (2004). [9] A. Tashakkori and J. Creswell, Exploring the nature of research questions in mixed methods research. Journal of Mixed Methods Research, 1(3), (2007) 207-211. [10] C. Teddlie and A. Tashakkori, Foundations of Mixed Methods Research: Integrating Quantitative and Qualitative Approaches in the Social and Behavioral Sciences. Thousand Oaks, CA: Sage, (2009). [11] R. Woodcock and M. Johnson, Woodcock-Johnson Psycho-Educational Battery-III. Chicago, IL: Riverside Publishing, (1989). [12] B. Glaser and A. Strauss, The discovery of grounded theory. Chicago: Aldine Press, (1967). [13] D. Mellard and M. Patterson, Contrasting adult literacy learners with and without specific learning disabilities. Remedial and Special Education, 29(3), (2008) 133-144. [14] B.D. Winn, C.H. Skinner, R. Oliver, A.D. Hale and M.R. Ziegler, The effects of listening-while-reading and repeated reading on the reading fluency of adult learners. Journal of Adolescent and Adult Literacy, (in press).