Implementation of Communicative Language Teaching (CLT) on educational robot for English language learning

English has become an important language for everyone to master. Therefore, innovative methods are needed to increase students' interest in learning English. Robot-Assisted Language Learning (RALL) uses robots to improve motivation in language learning. One of the applications is Communicative Language Teaching (CLT) using machine learning approaches such as Automatic Speech Recognition and Natural Language Processing (NLP). This paper proposes an educational robot designed for CLT implementation, which uses Speech-to-Text (STT), Text-to-Speech (TTS), and NLP to demonstrate the robot as a conversation partner. The robot has two training modes, which are training based on a scenario and training without using scenario. The training mode based on a scenario uses conversations that have been provided to train the user's ability to speak. The training mode without using scenario uses a Deep Learning-based NLP model trained using conversation datasets from Touchstone and Speak Now books. With a small number of pairs of question and answer, the NLP model can answer basic questions within the scope of the dataset but has a limited understanding of each vocabulary.