Uncertainty is a frequently occurring affective state that learners ex-perience during the acquisition of a second language. This state canconstitute both a learning opportunity and a source of learner frus-tration. An appropriate detection could therefore benefit the learn-ing process by reducing cognitive instability. In this study, we usea dyadic practice conversation between an adult second-languagelearner and a social robot to elicit events of uncertainty throughthe manipulation of the robot’s spoken utterances (increased lex-ical complexity or prosody modifications). The characteristics ofthese events are then used to analyze multi-party practice conver-sations between a robot and two learners. Classification models aretrained with multimodal features from annotated events of listener(un)certainty. We report the performance of our models on differentsettings, (sub)turn segments and multimodal inputs. ; QC 20200930 ; Collaborative Robot Assisted Language Learning
Detection of Listener Uncertainty in Robot-Led Second Language Conversation Practice
2020-01-01
Conference paper
Electronic Resource
English
Robot interaction styles for conversation practice in second language learning
BASE | 2020
|Robot interaction styles for conversation practice in second language learning
BASE | 2020
|Is a Wizard-of-Oz Required for Robot-Led Conversation Practice in a Second Language?
BASE | 2022
|