Engaging Test Takers in an English Speaking Task: Do Input Formats Matter?

ABSTRACT As multimodal input becomes integral to real‐world communication, test‐taker engagement has emerged as a critical concern in the design and validation of English speaking assessments. While previous research has examined performance outcomes across input types, few studies have investigated how input formats shape test‐taker engagement across multiple dimensions. This study addresses this gap by adopting a four‐dimensional engagement framework—behavioral, cognitive, social, and emotional—to examine how different input formats influence test scores and engagement in a speaking task based on a workplace negotiation scenario. Thirty‐eight Chinese undergraduates completed three versions of the task presented in audio, animation, and simulated‐live video formats. Engagement was measured using seven discourse‐based indicators capturing behavioral, cognitive, and social dimensions, complemented by questionnaire and interview data on emotional engagement. Results revealed no significant differences in test scores across formats. However, the simulated‐live video format, featuring a virtual interlocutor in a first‐person view, substantially enhanced cognitive, social, and emotional engagement. Notably, only social engagement (specifically direct address) significantly predicted test scores across all formats, whereas the other dimensions did not. These findings suggest that input format shapes how test takers interact with tasks, and further reveal differential links between engagement dimensions and performance. This study highlights the need for engagement‐sensitive task design, demonstrates the potential of simulated‐live tasks to foster more authentic and productive test‐taking experiences, and expands the operationalization of social engagement through pragmatic discourse features.