A topic-modeling assisted discourse analysis of undergraduate study abroad
The interest in text mining is recently increasing in the humanities and social sciences. Using a topic-modeling technique, this study analyzed a corpus of study abroad applications to explore a discursive field of study abroad. By doing so, this project finds the ways in which the new text analysis technique can contribute to the methodology of discourse analysis. For this purpose, 4,585 applications for a variety of undergraduate study-abroad programs were collected and sorted out into the corpora of successful and unsuccessful applications. The topic-modeling results show that generated topics generally match the discourses and themes that the existing research of study abroad have considered so far. The comparison of the results between successful and unsuccessful applications reveals that the former tends to exhibit a set of more clearly defined topics and use abstract and generalized words to describe actions engaging with study abroad. This study suggests that the topic-modeling technique can be a useful discourse-analytic tool as it helps understand a broad thematic and discursive terrain in a large size of textual data. This paper also discusses how traditional discourse analysis methods can contribute to addressing methodological limitations in text mining techniques.