Text-to-Speech as a Revision Tool? Well, Kind Of

Garrison, Kevin. “An Empirical Analysis of Using Text-to-Speech Software to Revise First-Year College Students’ Essays.” Computers and Composition 26.4 (2009): 288-301. Print.

In this article, Garrison explores the computerized version of reading essays aloud as part of the revision process. He says that text-to-speech (TTS) programs may provide some new possibilities in this regard. However, such technology had been underutilized and under-researched previously because of cost, ease of access, technological inadequacies (particularly the quality of the computerized voices), despite improvements in all three of these areas. Furthermore, most of the previous studies of TTS programs in written communication contexts focused on younger students (K-5) or students with disabilities and addressed how TTS programs helped these students address proofreading errors. Garrison’s study seeks to fill these gaps in experimental research on the effectiveness of TTS programs for global and local revisions of college-age students without disabilities.

His study employs a quasi-experimental design. He began with a pilot test with six students from one of his classes. After revisions, the study expanded to 52 students (again from his classes). The students were assigned to either a control or a test group. Those in the control group simply used Microsoft Word to revise while the test group used a TTS program. Once the test was over, he coded the data, comparing original and revised drafts. The results showed that the control group and the test group made about the same number of positive proofreading changes, but the control group made more positive global and local changes. The control group did make more neutral changes, suggesting the TTS program added some efficiency, but since the control group did better in terms of positive changes, the limited efficiency added by TTS programs seems to me a small reason to incorporate them (unless, of course, it meets a specific need of the students). But Garrison clearly notes that, based on his results, we cannot assume TTS programs to be a sort of revision magic wand. For TTS programs to be successful, they must be made a clear part of the pedagogy and their benefits/uses clearly explained to students before they are adopted.

Given the narrow scope of his study (52 students from his own classes), Garrison is appropriately cautious about his results. He neither rejects nor whole-heartedly endorses TTS programs. He was also careful in his methodology in many ways: for example, he conducted a pilot study and tested the reliability of his codes. But his sampling was maybe a little suspect. I wonder how free his students felt to volunteer or not for the study? He also discussed some concerns at the end of the study that he might have considered at the beginning. For instance, when he discusses the use of Microsoft Word for the control group, he explains students’ familiarity with the program to mitigate any concerns unfamiliarity might create. However, he gave the test group only a few minutes’ explanation of the TTS program and noted that their unfamiliarity with the program might have skewed some of the results. Such a variable might have been addressed in the preparation for the study.