Using Automatic Speech-to-Text Transcription in Language Sample Analysis

Language Sample Analysis is an important aspect of a comprehensive language assessment. It serves as a supplemental test, often utilized in addition to several standardized tests. With this being said, language sample analysis is not practiced regularly by the majority of Speech-Language Pathologists primarily due to barriers of time or efficiency, and lack of training in proper analysis or analysis programs. The purpose of this study is to identify whether time and efficiency can be addressed by automatic transcription via a speech-to-text program. Audio narrative language samples produced by typically developing school-age children will be used. Transcriptions will be constructed by Microsoft Dictate and compared to transcriptions created by trained transcribers. Utterance segmentation and word-level accuracy will be compared between the two transcript types. The results from this study will help to identify whether speech-to-text offers a means of reducing barriers to language sample use by Speech-Language Pathologists. 

Lucy Heller

Lucy Heller is a graduating senior at Saint Louis University from St. Louis, Missouri. She is majoring in Speech, Language, and Hearing Sciences and minoring in Spanish. This summer,  Lucy will begin graduate school at Saint Louis University where she will pursue her Master's degree in Speech-Language Pathology. 

Lucy is very thankful for the support given by Dr. Sara Steele through each step of this project. Dr. Steele has been incredibly influential in the development of Lucy's project through the duration of this school year. Her knowledge and guidance provided a strong foundation for learning and gaining a greater understanding of numerous aspects of language sample analysis, in addition to the research process as a whole.