"Cheese!" - a Corpus of Face-to-face French Interactions. A Case Study for Analyzing Smiling and Conversational Humor.

The aim of this article is to present Cheese! a conversational corpus and, as an illustration, an example of study that such a multimodal corpus allows. Cheese! consists of 11 French face-to-face conversations lasting around 15 minutes each. In this article, the methodology used to collect and enrich the corpus "Cheese!" is detailed: experimental protocol, technical choices, transcription, semi-automatic annotations, manual annotations of smiling and humor. Then, an exploratory study investigating the links between smiling and humor is proposed. Based on the analysis of two interactions of "Cheese!", two questions are asked: (1) Does smiling frame humor? (2) Does smiling have an impact on its success or failure? This study does not claim to fully answer these questions. Rather, it exemplifies which kinds of investigation are made possible by such a corpus. Although the experimental design of Cheese! has been elaborated specifically for the study of smiling and humor in conversations, the creation of high quality data set and the methodology presented can be replicated and applied to the analysis of many other conversational activities (such as narration or explanation) and other multimodal phenomena.