A Configurable Conversational Agent to Trigger Students’ Productive Dialogue: A Pilot Study in the CALL Domain

Abstract

Conversational agents constitute a specific type of ITSs that has been reportedly proven successful in helping students in one-to-one settings, while recently their impact has also been explored in computer-supported collaborative learning (CSCL). In this work, we present MentorChat, a dialogue-based system that employs a configurable and domain-independent conversational agent for triggering students' productive dialogue. After a system overview with an emphasis on design rationale and system architecture, we present a pilot study, where the agent is evaluated in the context of the computer-assisted language learning (CALL) domain. Thirty students collaborated in small groups trying to accomplish three different tasks. MentorChat conversational agent supported each group discussion differently in each task providing (a) 'weak'-directed interventions and/or (b) undirected interventions. The study results indicate that 'weak'-directed agent interventions can be more effective than undirected interventions by means of increasing the level of explicit reasoning, and thus productive dialogue. Some encouraging results concerning the system usability and user acceptance are also presented.

Research on e-Learning and ICT in Education

“…(Veletsianos & Russell, 2014 ) Following this potentially promising research direction, we have argued that conversational agents for collaborative learning can be designed by focusing on the role of the teacher as well as the peers' interactions occurring while students work together (Tegos, Demetriadis, & Tsiatsos, 2012 ). Based on this rationale, we have developed a prototype conversational agent system, named MentorChat (Tegos, Demetriadis, & Tsiatsos, 2014 ). In the following sections, we present an overview of the MentorChat system and an evaluation study exploring how the students' perceptions of the agent and their conversational behavior may be affected by the different roles (peer or tutor) of a conversational agent.…”

“…MentorChat is a cloud-based multimodal dialogue system that utilizes an embodied conversational agent to scaffold learners' discussions (Tegos et al, 2014 ). We have developed MentorChat as a domain-independent dialogue system that (a) promotes constructive peer interactions using facilitative agent interventions (prompts) and (b) enables the teacher to confi gure the support provided by the conversational agent.…”

Why not Robot Teachers: Artificial Intelligence for Addressing Teacher Shortage

“…Designing classroom-based robotic personalities, and especially those that can take on teacher roles, is however challenging (Sharkey 2016). Studies within education that have reported on the roles of robots are many and diverse, and robots in education have been noted to address absenteeism (Iver, Abele, and Douglas 2014), acting as triggers for productive dialogue in language instruction (Tegos, Demetriadis, and Tsiatsos 2014), provided emotional support for learners (Dennis, Masthoff, and Mellish 2016), and promoting creativity and problem-solving (Liu et al 2013). The implications for students' future wellbeing has also been discussed (Saltinski and Ph 2015).…”

Conversational Agent to Promote Students' Productive Talk: The Effect of Solicited vs. Unsolicited Agent Intervention

“…[3] Should the agent address the whole group or just a specific group member in order to efficiently scaffold students' online discussions? [11] Another intriguing research question is whether the supportive interventions of the agents should be offered automatically or only after students accept an agent invitation for support. Chaudhuri et al [5] addressed a similar research question in a classroom study.…”

“…More specifically, we calculated the 'explicit response ratio' (ERR) for both the solicited and the unsolicited intervention mode. The ERR, which was introduced and used in our previous study [11], constitutes a statistical measure of how many explicit responses are expected to be triggered by an agent intervention. Thus, ERR can range from zero, if no explicit responses are elicited, to more than one, if multiple responses are induced by an agent intervention.…”