The ability to understand stances adopted by participants during interactions is an important aspect of human social intelligence. Stances can be seen as expressions of attitudes, feelings, judgments, or evaluations concerning the propositional content of a message, (i.e., statements, assertions, or verbal and
non-verbal behaviors which express judgments or opinions) that sometimes have an affective component. One can say, for example, that someone is in a calm, assertive or uncertain stance. They contrast
with personality traits and emotions, in that stances are rather stable but not permanent states, they are adopted rather than elicited, they are addressed to someone or something, and are
co-created by the participants during interaction. Furthermore, stance expressions are multimodal in nature and can involve linguistic, paralinguistic, facial and bodily behavior. Understanding stances
during interactions allows humans to have fluent and successful conversations. One of the core aims of social signal processing (SSP) is to carry this human skill over to machines.

Allowing machines such as embodied conversational agents (ECAs) to have these human-like interactions with humans involves automatically recognizing and synthesizing stance behavior. The recognition of a user’s stances during interaction will enable an ECA to identify how
the user’s message is to be perceived and understood, and consequently, how the interaction can be adapted to this
interpretation. The synthesis of an ECA’s stances improves the interaction in terms of, for example, the ECA’s believability and the atmosphere of the interaction.

Hence, this workshop concentrates on studies of multimodal expressions of stances in both human-human and human-machine interaction, and on the automatic processing of stances. We aim to bring together researchers from various disciplines (e.g., phonetics, linguistics, psychology, computer science) to foster multidisciplinary discussions on stances in interaction.