Garrett, J. T., Heller, K. W., Fowler, L. P., Alberto, P. A., Fredrick, L. D., & O’Rourke, C. M. (2011). Using speech recognition software to increase writing fluency for individuals with physical disabilities . Journal of Special Education Technology , 26 (1), 25–42. https://doi.org/10.1177/016264341102600104

Speech recognition software was examined in comparison with word processing—without spelling or grammar checks, auto-complete and auto-correct features—for composing written products rather than handwriting. Additional individualized supports included hardware such as roller ball mouse, eyeglasses, 16-point text enlargement, mini keyboard, and pen for typing.

Five (5) high school students with physical/orthopedic disabilities impairing use of their hands for fine motor tasks, apparently from Georgia (U.S.), participated. Additional details of the disabilities included that two students had spina bifida, one had spinal muscular atrophy, one had cerebral palsy along with visual impairment and Asperger syndrome, and one had Duchenne muscular dystrophy. Participants included one in grade 9, two in grade 10, and two in grade 12; their ages were also reported. Nearly all students were in general education, and one was in a self-contained placement. Participants' amount of experience with using a word processing program was individually documented, as was their experience with speech recognition software.

To gather baseline data for the study the Wide Range Achievement Test, Third Edition (WRAT-3; Wilkinson, 1993) measured spelling skills and Microsoft Word's Flesch-Kincaid readability scale measured the grade level of their written compositions. Participants' handwriting and typing rates were measured in word count per minute. Participants were next provided training on the speech recognition software. Each participant was provided word processing software and speech recognition software, in an alternating treatment design, for composing essay drafts for three minute sessions. Composition samples were analyzed for fluency—speed in word count per minute—for both word-processed and verbalized products; accuracy was also measured, including word errors. At the conclusion of the writing activity, participants were given a questionnaire for relaying their perceptions of the speech recognition software accommodation.

All five participants in the study demonstrated greater writing fluency when using the speech recognition software in comparison to word processing, either immediately for four participants, or over the course of composing the series of writing samples for one participant. When using speech recognition, participants' composition lengths averaged longer than when using word processing by 57, 80, 91, 100, and 226 total words for the three-minute sessions. The participants produced samples with total words averaging 88, 62, 23, 25, and 88 words when using word processing. Participants tended to receive higher accuracy scores for writing samples produced using word processing, averaging higher scores—than when using speech recognition—of 4%, 6%, 11%, 14%, and 21%. Participants also tended to recall the types of errors they made more successfully when using word processing than when using speech recognition. Participants all indicated liking to use the speech recognition accommodation, and that they worked faster than word processing. Four of the five perceived using speech recognition as easy or very easy. One student preferred using word processing, two students preferred speech recognition, and the remaining students indicated liking both; three indicated that they planned to use speech recognition only, and one planned to use both.