What Code-Switching Strategies are Effective in Dialogue Systems?

Since most people in the world today are multilingual, code-switching is ubiquitous in spoken and written interactions. Paving the way for future adaptive, multilingual conversational agents, we incorporate linguistically-motivated strategies of code-switching into a rule-based goal-oriented dialogue system. We collect and release CommonAmigos, a corpus of 587 human-computer text conversations between our dialogue system and human users in mixed Spanish and English. From this new corpus, we analyze the amount of elicited code-switching, preferred patterns of user code-switching, and the impact of user demographics on code-switching. Based on these exploratory findings, we give recommendations for future effective code-switching dialogue systems, highlighting user's language proficiency and gender as critical considerations.