Exploring the validity of voice recognition for formative classroom-based assessment: An oral reading fluency application
Darla K. Jones, University of Oregon, United States
University of Oregon . Awarded
This study employed a descriptive exploratory case study design to evaluate the accuracy and efficiency of using voice recognition technology to collect oral reading fluency (ORF) data for classroom-based assessments. The primary research question was as follows: Is voice recognition technology a valid and reliable alternative to traditional oral reading fluency data collection methods? Both quantitative and qualitative data were collected and analyzed to address this question.
Fourteen students from a single fourth-grade class participated in the study. Ten sessions were conducted in which 2 ORF samples were collected for each student: one traditional, and one using VR technology, resulting in a total of 20 samples per subject. For the purposes of this study, the 95% accuracy level was designated as the minimum acceptable standard for content validity. Accuracy rates were then studied both graphically and through statistical analysis, utilizing single factor ANOVA for repeated measures and the Bonferroni/Dunn procedure for post hoc comparison of session pairs. Twenty session pairs showed a significant difference in accuracy rates (p = .0011).
Although 11 of the 14 students participating in the study had at least one session in which a 95% accuracy rate was achieved, results from this study were not consistently high enough to be considered valid or reliable for assessment purposes. Results indicated that the VR data collection process employed in this study was clearly and significantly less accurate than traditional data collection methods.
Nonetheless, accuracy rates were fairly stable over time, indicating that voice recognition should be further investigated as a potential assessment tool, especially if more time were invested to improve VR accuracy rates. Findings also indicated that student accents, reading speed, colds or allergies, and background noise were not major factors influencing accuracy rates for this study. Only one factor (a speech impediment in the form of a lisp) was a major barrier to accuracy, resulting in the exclusion of one subject from the sample. The study concludes with specific recommendations for voice recognition implementation and suggestions for future research.
Jones, D.K. Exploring the validity of voice recognition for formative classroom-based assessment: An oral reading fluency application. Ph.D. thesis, University of Oregon.
Citation reproduced with permission of ProQuest LLC.
For copies of dissertations and theses: (800) 521-0600/(734) 761-4700 or https://dissexpress.umi.com