Compensatory Effectiveness of Speech Recognition on the Written Composition Performance of Postsecondary Students with Learning Disabilities

Abstract

Seventeen males and twelve females wrote essays under three conditions: (a) without assistance; (b) using a human transcriber; and (c) using a speech recognition system. Students received higher holistic scores using speech recognition than when writing without assistance at a statistically significant level ( p=.048). In order to determine the reasons for the superior scores on the essays written using speech recognition, 22 measures of fluency, vocabulary and syntax were computed. Several measures showed a strong correlation with the holistic score. A multiple regression revealed that the best predictor of the holistic score was Words with Seven or More Letters. Further, the ratio of Words with Seven or More Letters to Total Words differed significantly across conditions ( p=.0136), in favor of speech recognition, when compared with receiving no assistance. A factor analysis identified three factors that accounted for significant variation in holistic score: Factor 1, measures related to length of the essay ( p=.0001+); Factor 2, measures of morphological complexity ( p=.003); and Factor 3, main verbs ( p=.021). The authors suggest that the technology may have been successful because it “encouraged” the use of longer words, a powerful predictor of a holistic score.