Investigating predictability aspects of phonological type errors in braille spelling

Abstract

BackgroundΤhe present study focuses on mechanical processes of braille writing conducted by the fingers. Analyzing and attempting to understand the perceptive feedback of the fingers is in specific situations important to develop methods to minimize phonological type errors. AimDuring a braille spelling task, the authors examined the phonological errors and traced the most and the least error prone finger(s). The authors also examined if it was possible to predict the type of phonological errors in braille writing based on four independent variables relevant to braille code. MethodThe error rate of twenty nine participants was estimated by normalizing the frequency of errors performed by each finger in relation to the frequency of the executed keystrokes. Additionally, a multinomial logistic regression examined the potential effect of the column of the braille cell, the dot density of the braille characters and the word length on the type of phonological errors, while a chi-square test has been performed between types of error and fingers. ResultsThe fingers were not equally error prone to phonological type errors. The multinomial logistic regression analysis revealed that when a replacement error appeared, the odds for an omission error were significantly less in the right column as well as in short words, while the probability for an addition error was significantly less in both cell columns. In contrast, when a replacement error appeared, the participants tended to perform more omission errors in words with rare density. ConclusionsThe right ring finger was the most error prone, while the right middle finger was the least error prone. The multivariate logistic regression model showed that all the independent variables, particularly the cell column, exhibited a statistically significant relation with the types of errors.