Measuring Similarities Between External and Self Emotion Evaluation in the Case of Assisted Speech Therapy of Children

Abstract

The emotional state of children involved in speech therapy represents an important condition for a successful intervention and, in consequence, the lack of emotional assessment is one of the most important drawbacks of computer assisted speech therapy. In this paper we focus on the second step of development of an automatic emotion recognition framework, that is identifying the consistency of human emotion recognition. Our research hypotheses were tested using results from an experiment involving 41 children, 3 speech and language therapists (SLT) and 3 therapeutically sequences. We developed and used an instrument to obtain data about children emotions from SLT (external emotion evaluation) and from the involved children (self emotion evaluation). Data were processed by two different mathematical methods for checking the similarity. The results have shown that there is a good consistency between external and self emotional evaluations and these findings encourage us to continue our researches for adding an affective extension to our Computer Based Speech Therapy (CBST