What do Filipino speakers talk about when they switch to Tagalog? A Topic Modeling Approach

Alianda Lopez

Tagalog-English code-switching is a common occurrence in the Philippines, and one that still warrants further research. This study sheds light on this phenomenon by employing a topic modeling approach to identify topics where code-switching occurred in speeches given in academic events. This is particularly interesting as this is the first study to use a formal and planned context to investigate code-switching, as most use data from spontaneous and oral conversations. This study is also one of the few studies in that employed a text mining approach on data containing Philippine languages. A Latent Dirichlet Allocation (LDA) model was built to generate topics from portions of academic speeches that contained Tagalog lexical items. Six topics were identified, namely, achieving goals, gratitude for companionship, education, staying true to values, greater good, and making a difference. Results showed that the topic on gratitude for companionship was found in the most number of speeches, while achieving goals occurred most frequently as the dominant topic. This shows that occurrences of Tagalog-English code-switching differ depending on the kind of context involved. The results merit further study to look into the differences found.