International Conference on Computer and Knowledge Engineering

Home / 11th International Conference on Computer and Knowledge Engineering

An intelligent linguistic error detection approach to automated diagnosis of Dyslexia disorder in Persian speaking children

Authors :

Fatemeh Asghari¹ Mahsa Khorasani² Mohsen Kahani³ Seyed Amir Amin Yazdi⁴ Mahdi Arkhodi Ghalenoei⁵

1- Department of Computer Engineering Ferdowsi University of Mashhad Mashhad, Iran 2- Department of Computer Engineering Ferdowsi University of Mashhad Mashhad, Iran 3- Department of Computer Engineering Ferdowsi University of Mashhad Mashhad, Iran 4- Department of Counseling and Educational Psychology Ferdowsi University of Mashhad Mashhad, Iran 5- Department of Counseling and Educational Psychology Ferdowsi University of Mashhad Mashhad, Iran

Keywords :

Automatic diagnosis, Computational linguistics, Deep learning methods, Linguistic error detection, Mental health disorders, Dyslexia disorder, Persian children

Abstract :

Dyslexia is a learning disability in which a child with a normal IQ has difficulties with reading. Each of these difficulties is linked to a certain type of weakness, such as visual memory impairment, auditory sensitivity impairment, attention, and so on. If left undiagnosed, the disorder grows with the child's development and, due to insufficient awareness of the parents, teachers and other people who interact with him / her, causes problems such as frustration and feelings of weakness in comparison to other children. Therefore, the need for early detection of this disorder at a young age is very significant. Educational scientists working in the field of diagnosis and treatment of learning disabilities use various standardized tests for screening. However, the use of intelligent systems for automatic diagnosis of dyslexia can be performed for initial screening, on a large scale and with less time costs and specialized manpower. In this study, using computational linguistic methods, we extract differential features of dyslexia from linguistic samples of Persian children with and without Dyslexia disorder, to train a machine learning model for automatic diagnosis of the disorder. The proposed method is leading the performance at classifying Dyslexic and non-Dyslexic Persian children with 0.94 and 0.95 for precision and recall measures, for trained classification models with Multilayer Perceptron and Decision Tree algorithms, respectively.