Speech is a fundamental aspect of human communication, produced through a coordinated effort between the larynx and the pulsating airflow from the lungs. Speech Sound Disorders (SSDs) in children encompass a broad spectrum of difficulties in producing speech sounds that adversely affect intelligibility and communication. In the United States, approximately 11 percent of children aged three to six experience a speech disorder, posing significant challenges for early detection and intervention. Traditional auditory-perceptual assessments and manual transcription methods are time-consuming and prone to subjective errors and biases, especially given the inherent phonetic variability in children’s speech. Although Automatic Speech Disorder Detection (ASDD) systems have been widely studied and applied in adult populations, their adaptation for pediatric applications remains underexplored. This study investigates the effectiveness of self-supervised learning (SSL) representations within an ASDD framework specifically designed for young children. Three popular SSL models, pre-trained with different configurations, are utilized, and performance is compared across various SSL model layers. Evaluations also contrast these representations with standard hand-crafted features, highlighting diagnostic accuracy and efficiency improvements. The results demonstrate that SSL-enhanced ASDD systems can offer a promising solution for the early detection of speech disorders in children, addressing a critical gap in current research and paving the way for more reliable, automated diagnostic tools. This work establishes a foundation for future innovations in pediatric speech disorder diagnostics.