Quantitative analysis of qualitative data: using text mining techniques for the music therapy clinic

Abstract

Text mining is a data mining process that aims to extract useful information from unstructured or semi-structured data sets, such as emails, HTML files, or transcripts. This study proposes the use of text mining techniques to evaluate verbal communication in music therapy sessions. To this end, sessions conducted by the project Effects of Music Therapy Interventions on the Quality of Life of Black Women were transcribed. The techniques of word cloud, topic modeling, and sentiment analysis were employed in this research. Topic modeling identified four clusters of words named "Black Women", "Reflections", "Music Therapy" and "Pain". In turn, sentiment analysis identified a broad distribution of sentiments, with a predominance of neutral or positive sentences, strongly skewed to the right and concentrated around the mean, with few extremely negative or positive sentences. These data suggest that the sessions addressed sentimental issues related to the Black women participants and promoted well-being for this population. According to the participants' reports in response to a questionnaire at the end of the music therapy sessions, the results of the text mining are in line with the participants' perceptions. In this sense, text mining is a promising technique for understanding verbal communication in music therapy sessions, and can be used in future research to investigate, with more evidence, these and other possibilities for music therapy clinical practice.