Speech disorders in children are a condition that could reduce the opportunity to access education, health care and in the future could mean a worse socioeconomic outcome. Therefore, early diagnosis and timely therapy is really important to reduce their impact in later stages of life. This paper presents a method for the gathering of data for a corpus related to Speech Disorders in children; such corpus will serve as the base to generate a semi-automatic ontology intended as a tool for therapists to help in the diagnosis and shape up of a therapy strategy.