Who, when and why: The 3 ws of code-switching

Abstract

With the rise of globalization, the use of mixed languages in daily conversations, referred to as “code-switching” (CS) has become a common linguistic phenomenon among bilingual/multilingual communities. It has become common for people to alternate between distinct languages or “codes” in daily conversations. This has placed a high demand on Natural Language Processing (NLP) applications to be able to deal with such mixed-language input. Researchers have lately achieved advancements in multilingual NLP applications, however, few work has been done to adapt these applications to users’ CS behaviour. In this work, we take the first steps towards this goal by investigating the CS behavior and its correlation with the users’ profiles in our case study on Egyptian Arabic-English code-switching. Although these factors have been investigated by linguists, the findings have been mostly made through theoretical studies. We provide empirical evidence based on a user study with 50 participants showing initial correlations between user traits and the CS frequency, which can be used to predict users’ CS. Our findings imply that in the scope of our study, people (who) code-switch in specific discourse domains more than others (when) and depending on their background and social factors (why). The study also shows that to be able to properly investigate who code-switches more data needs to be collected and further analysis is required.