Willingness to communicate (WTC) in a second language class has been shown to be transient and situation‐dependent. Previous studies mostly elicited data from learners' self‐reports through interviews or diaries, and the data analyses were largely confined to the medium of language, marginalizing other semiotic resources such as gesture or movement. The nature of human communication, however, is multimodal, which entails the processing of different semiotic resources. Drawing on a systemic functional multimodal discourse analysis approach, this study investigates the dynamics of WTC in the English as a foreign language classroom in China. The researchers collected data through videotaping English lessons, stimulated recall interviews, and learning journals. Their analyses involved two teaching scenarios videotaped in one class period, which were rated respectively with high WTC and low WTC by four participating students in stimulated recall interviews. These scenarios were transcribed and annotated using Multimodal Analysis Video software to identify the use of semiotic resources, including language, gesture, and gaze. The discourse semantic features of the two scenarios were analyzed and compared, and subtle differences were presented and discussed. This study's findings highlight the need for language teachers to recognize and orchestrate multimodal semiotic resources to enhance students' WTC and classroom participation.