Is Language Use with a Virtual Interlocutor Different than with a Human?

As humans, we increasingly interact with non-human interlocutors such as smart speakers, voice assistants, and virtual agents. To what extent is our language use with such virtual interlocutors similar or different to the way we communicate with humans? In a cross-disciplinary scoping review, we find that humans use language with virtual interlocutors in a similar manner as they do with humans, particularly when virtual interlocutors (visually) appear human-like. Sometimes there is more lexical alignment and less phonetic alignment with virtual interlocutors than with humans, arguably in an attempt to facilitate communication (through lexical repetition and hyperarticulation) with a virtual addressee that is deemed less competent than the average human interlocutor. These findings have important theoretical implications for our understanding of human communication and provide practical insights for the societal optimization of communication between ourselves and the virtual interlocutors that increasingly surround us.