The Computational Cultural Understanding (CCU) program aims to create human language technologies that will help to identify cross-cultural misunderstandings in dialog that may lead to conversational failures. Cross-cultural studies by psychologists, ethnographers, and anthropologists have provided much evidence that emotions differ in different cultures in multiple ways, including how they are defined and how (or whether) they are expressed. Conversational and social norms also differ across cultures. Our Cross-Cultural Harmony though Affect and Response Mediation (CHARM) team focuses on developing multimodal, cross-lingual, and cross-cultural models for sociocultural norms discovery, cross-cultural emotion recognition, and detection of impactful changes in communication that may lead to either failure or success of the overall conversation.