Prosodic skills may be powerful to improve the communication of individuals with intellectual and developmental disabilities. Yet, the development of technological resources that consider these skills has received little attention. One reason that explains this gap is the difficulty of including an automatic assessment of prosody that considers the high number of variables and heterogeneity of such individuals. In this work, we propose an approach to predict prosodic quality that will serve as a baseline for future work. A therapist and an expert in prosody judged the prosodic appropriateness of individuals with Down syndrome' speech samples collected with a video game. The judgments of the expert were used to train an automatic classifier that predicts the quality by using acoustic information extracted from the corpus. The best results were obtained with an SVM classifier, with a classification rate of 79.30%. The difficulty of the task is evidenced by the high inter-human rater disagreement, justified by the speakers’ heterogeneity and the evaluation conditions. Although only 10% of the oral productions judged as correct by the referees were classified as incorrect by the automatic classifier, a specific analysis with bigger corpora and reference recordings of people with typical development is necessary.