Applying situation-awareness for recommending phonological processes in the children's speech

Situation-Awareness (SA) involves the correct interpretation of scenarios, allowing a system to respond to the observed environment in several domains. Speech therapy is an area where SA may provide benefits; however, the related literature generally is not concerned with identifying phonological processes (PPs) in pronunciation and their effects on the management of therapeutic tasks. An early identification of speech sound disorders allows the diagnosis and treatment of various pathologies and the reasoning about situations may aid clinical decision-making. So, in this paper, we present a novel method for predicting PPs, supporting speech therapists in the identification of speech disorders in children. Our approach uses SA tied to machine learning to first classify the correctness in the pronunciation of a set of target words. Then, a second instance of ML uses scores calculated from mispelled words to predict the PPs. The method was evaluated through a speech corpus containing over a thousand of audio files, collected from pronunciation assessments performed by speech-language pathologists with more than 1,000 children. Our results showed an average accuracy over 92.5% for classifying the pronunciations, and 92.2% for predicting the PPs.