Qilin, a Robot-Assisted Chinese Language Learning Bilingual Chatbot

Robot-assisted language learning (RALL) offers a potentially fruitful approach to technology-enhanced instruction especially during the pandemic, as robot-learner interaction enhances a more immersive and personalized learning environment. However, the effectiveness of RALL within the context of an English speaker learning a tonal language such as Mandarin Chinese has been minimally researched. This manuscript presents a quasi-experimental study on the comparison of a multimodal virtual agent and embodied robot in influencing an adult English speaker's ability and interest in learning spoken Mandarin Chinese by carrying out a dialogue-based language learning activity. Half of the study participants were placed in a “control” group, tested with a completely virtual agent. The remainder were placed within an “experimental” group, tested with a physical robot. Participants interacted with three modes: translation, chat, and quiz of the bilingual chatbot Qilin. This was followed by the completion of a survey with 10-point rating scales and a qualitative question detailing their experience. Qilin utilizes speech services offered by the Google Cloud platform along with visual, auditory, and kinetic indicators according to respective operative agents. Favorability towards the physical embodiment of the robot resulted in higher levels of engagement and lower levels of perceived discouragement when interacting with the physical agent even though the speech and audio functions were identical for the virtual agent. The nonnative tones of the speakers and the polyphonic characters of Mandarin Chinese primarily contributed to levels of discouragement. An analysis of initial results and suggestions of future works is provided to foster research in Chinese RALL for English speakers.