Noakes, M. A. (2017). Does speech-to-text assistive technology improve the written expression of students with traumatic brain injury ? (Publication No. 10602238) [Doctoral dissertation, Duquesne University]. ProQuest Dissertations and Theses Global. https://www.proquest.com/docview/1954694875
Duquesne University (Pittsburgh, PA); ProQuest document ID: 1954694875; also located on Duquesne University at https://dsc.duq.edu/etd/181
Dictated response using speech recognition technology (specifically Dragon Naturally Speaking) was compared with compositions produced with standard handwriting. The researcher noted that individual administration was an additional aspect of testing permitted by virtue of the students' needing to speak their responses aloud; students also completed handwritten compositions in an individual administration and quiet space. The study provided several practice sessions and repeated data collection to analyze for multiple comparisons between handwritten and dictated response across the 5 sessions in 5 weeks. Participants produced compositions with the dictated response accommodation and with handwriting in each of the five weekly sessions.
Three students with traumatic brain injuries (TBIs), one each in grades 4, 8, and 9, participated in this alternating treatments design experiment. Additional information, such as gender and ethnicity, as well as the timing of their acquired conditions and the individual expressions of their TBIs, was reported.
Participants were prompted to construct narratives using ten different story prompts throughout the study. The Written Expression Curriculum Based Measurement (WE-CBM) tracked participants' progress in correct written sequences (CWS), total words written (TWW), and words spelled correctly (WSC). These three elements were quantified by the WE-CBM into separate scores. As noted in the study's Abstract, the focus of providing additional support for students with TBIs was to improve the "quality, construction, and duration of their written expression." The researcher described the study as an alternating treatments design experiment, analyzing data for incidence of non-overlapping data points, and reporting effect sizes for significant difference in performance between testing conditions. Additionally, observations of students' actions while completing their writing composition tasks were also reported.
All three participants demonstrated significantly higher written performance when using the dictated response accommodation, employing speech recognition software on a digital platform (computer). The researcher reported large effect sizes, ranging from +3.4 to +8.8. Further, the researcher noted "Perhaps more impressive, was 100 percent non-overlap of data between the two conditions across participants and dependent variables" (p. 67). Observations of participants' behavior and handwritten products yielded detailed examples to support the CBM data indicating that their standard unaccommodated performance was substantially limited by their disabilities; in contrast, the speech-to-text accommodation supported "significantly improved performance" (p. 69). The researcher indicated that the three participants benefited despite their traumatic brain injuries and related fine motor skill impairments. She suggested that the results supported the intent of addressing the needs of students with TBIs by removing the barrier to performance, noting the mechanism believed to be involved: "In theory, because less effort is spent on transcription, there is a reduction in cognitive load, enabling more time to be spent on generation skills, such as idea development, selecting more complex words that might be otherwise difficult to spell, and grammar" (p. 66). Limitations of the study were reported, and future research directions were suggested.