The Interactive and Situated Character of Daily Conversation
In Conversation Analysis (CA), we use the terms ‘interactive talk’ or ‘talk-in-interaction’ to refer to a particular type of spontaneous everyday conversation. Interactive talk doesn’t just involve two people taking turns to say something, with the listener paying attention only to the verbal content of what is said. Rather, interactive talk is orderly and situated communication. It draws its meaning from the way speech is structured and from the context in which it is produced, as well as from the words that are said. It relies on a common set of social and interactive practices shared by members of a speech community, a set of rules that allows meaning to be conveyed in the way that words are spoken as well as by the words themselves.
Interactive talk involves the transfer of underlying thought structures (tacit reasoning) into ‘natural talk’, or normal conversation. The ability to communicate the universally valid thought structures shared by all humans is critical for maintaining the possibility of communication between different, unique and distinct subjects. At the same time, defining a message as ‘situated’ means that its understandability is intimately linked to the contextual framework in which it has been produced.
Virtual Assistants’ Interactive and Situated Talk
In Artificial Intelligence, the term ‘situation’ describes the way in which a robot or a software agent is embedded in an environment. Virtual assistants are situated in dynamic contexts that are subject to change and variation. And they aren’t just passive parts of the background furniture; they can manipulate, change, and perceive their contexts by interacting with them. As such, when we want to optimise the conversational practices of a virtual assistant, we need to go beyond a simplistic analysis of the meanings of the words they use. Instead, we need to pay close attention to the context in which they use them. In linguistics, this contextually grounded approach is called ‘pragmatics’.
Virtual assistants’ conversational practices can be improved by analysing them from…