Young, M. C. (2017). The effects of text-to-speech on reading comprehension of students with learning disabilities (Publication No. 10254709) [Doctoral dissertation, Illinois State University]. ProQuest Dissertations and Theses Global. https://www.proquest.com/docview/1927469340

Illinois State University (Normal, IL); ProQuest document ID: 1927469340; accessible on Illinois State U's website at https://ir.library.illinoisstate.edu/etd/718/

Text-to-speech was provided using Kurzweil 3000 on the computers during performance task completion. Participants pre-selected their settings, including highlighting, rate of speed, and voice selection. Planned task periods—5 baseline and 7 intervention sessions, along with maintenance sessions, in an A-B-A-B withdrawal design—using and not using text-to-speech for each participant were documented, and performance data effects were compared.

Four students in grade 9 who had learning disabilities related to reading completed various performance tasks while receiving and not receiving the text-to-speech accommodation. One student also had attention-deficit hyperactivity disorder (ADHD), and another also had autism. Demographic information such as gender, age, ethnicity, and socioeconomic status (qualifying for free and reduced lunch), were also reported for the students, and for the high school's student population as a whole, as well as for the special education population. The setting was a large Midwest (U.S.) public high school. All four students received English instruction in self-contained classroom, along with up to 11 other students; both participants and non-participants in the class completed similar individual performance tasks at the same time during the class periods. The proportion of time in general education was also reported for each participant.

A curriculum-based measure (CBM) of 15 multiple choice items was presented to students, with eight vocabulary items, four literacy analysis items, and three reading comprehension items. Oral reading fluency (ORF) was also assessed, with participants reading aloud to the test administrator in a baseline pre-test; ORF was checked after each accommodation condition, as well as in a post-test. The ORF pre-test provided information for setting the text-to-speech rate of speed. The Scholastic Reading Inventory (SRI) served as a separate reading comprehension assessment, measured in Lexile scores. General performance data—including levels and trends within conditions, and levels between conditions—and percentage of nonoverlapping data (PND) were analyzed for comparison between accommodated and nonaccommodated conditions. Participants also completed an 8-item survey including a rating scale about their perceptions as a social validity check.

Three of the four participants demonstrated significant positive differences in CBM scores between the unaccommodated condition and the oral delivery accommodation—using text-to-speech. Individual participant scoring patterns were also reported. Vincent demonstrated limited average improvement in the overall CBM score, with his smallest improvement in the comprehension component, and lower than typical improvement (compared to age-mates) in oral reading fluency (ORF) from the baseline to post-intervention [he missed the maintenance sessions]. While there were increases in some of Vincent's scores, particularly in abrupt increases in immediacy of effect, they were not significant overall. Jack showed significant mean improvement from first baseline (34%) to maintenance (75%) scores, with his largest improvement in the vocabulary component. He also demonstrated significant increases in immediacy of effect, and scores showed accelerating and therapeutic trends. Jack also had a 100% in PND (percentage of nonoverlapping data). With the lowest (of the four participants) initial ORF (that is, words per minute) score, Jack made the largest improvements—in fact, he made twice the growth rate in comparison to average peers. Dianna's overall CBM scores also showed a significant immediate effect from using text-to-speech; her scores had a similar significant decrease when switching from the accommodated to unaccommodated conditions; her largest improvement was in comprehension component average: first baseline of 33% to maintenance of 67%. Her scores showed an accelerating growth pattern, and reached 100% PND. Dianna's ORF scores were significantly higher overall, associated with text-to-speech. Donald showed an overall mean CBM increase from 44% at baseline to 68% at maintenance, deemed a significant effect from text-to-speech, with similar improvements in vocabulary, literary analysis, and comprehension components. His scores showed accelerating and therapeutic gains, and reached 100% PND. He demonstrated an average increase in ORF scores. Two of the four students—Jack and Dianna—showed improvements in comprehension on a screening assessment, SRI, indicating the ability to generalize this skill. The survey results indicated that the social validity of using text to speech for all three students (with one not completing this survey) was positive overall at the end of the study. Specifically, the most positively rated statements indicated enjoyment—highest, at 4 out of 5—and easy-to-use, remembering more information, and liking to moderate the speed of speech (all of which had the mean rating 3.7 of 5). Limitations of the study were reported, and future research directions were suggested.