A hybrid approach to assessing spoken fluency combining three metrics with support vector machines

Utilizing Pearson-r correlations and Support Vector Machine (SVM) analyses, this paper provides specific evidence regarding the extent to which the interface between human and computer evaluations of spontaneous engaged speech provide statistically significant measures of fluency. Three types of measures are used: quantitative measures, common sense notional measures, and comprehensive measures. As such, it contributes to the growing body of literature describing the current limits of automatic systems for evaluating spoken proficiency, it supports the continued development and implementation of hybrid systems, and it includes suggestions for the utilization of additional automatic analyses within a hybrid system.