Voice user interface has become a commercially viable and extensive interaction mechanism with the development of voice assistants. Despite the popularity of voice assistants, the academic community does not utterly understand about what, when, and how users chat with them. Chatting with a voice assistant is crucial as it defines how a user will seek the help of the assistant in the future. This study aims to cover the essence and construct of conversational AI, to develop a classification method to deal with user utterances, and, most importantly, to understand about what, when, and how Chinese users chat with voice assistants. We collected user utterances from the real conventional database of a commercial voice assistant, NetEase Sing in China. We also identified different utterance categories on the basis of previous studies and real usage conditions and annotated the utterances with 17 labels. Furthermore, we found that the three top reasons for the usage of voice assistants in China are the following: (1) greeting, (2) function, and (3) music. Chinese users like to interact with voice assistants at night from 7 PM to 10 PM, and they are polite toward the assistants. The whole percentage of negative feedback utterances is less than 6%, which is considerably low. These findings appear to be useful in voice interaction designs for intelligent hardware.

본 연구는 인공지능 기술과 메신저용 챗봇의 인터페이스 융합을 통해 의사소통능력 개발을 위한 외국어 학습의 효과를 검증하는 목적으로 수행되었다. 이를 위해 첫째, 자체 학습이 가능한 로봇 기반과 인공지능 기술과 디지털 융합기술을 적용한 학습시스템을 설계하여 학습 성취도에 미치는 영향을 조사하였다. 둘째, 휴머노이드 로봇의 인간과의 상호작용 구현 능력을 검증하고 인터페이스 프로그램을 자기학습의 알고리즘으로 의사소통 능력 학습에 적용하여 그 효과를 검증하고자 하였다. 셋째, 아이팟, 아이폰, 아이패드, 매킨토시PC, iTV 등 모든 기기에 아이튠즈와 앱스토어를 탑재하고 있는 클라우드 앱과 페이스북 메신저 챗봇의 다양한 콘텐츠를 학습시스템에 활용하였다. 연구 대상은 충남 천안시 대학에서 재학 중인 학생 120명이었다. 실험집단과 통제집단의 사전 시험, 사후 시험, 학습 활동에 대한 평가를 통한 성취도를 분석하고 또한 실험집단의 설문 조사를 통하여 통계분석을 하였다. 연구결과는 다음과 같다. 첫째, 로봇은 학생과 같은 콘텐츠 사용자, 교사 같은 콘텐츠 제작자, 실제 자료 콘텐츠 간의 상호작용을 강화할 수 있어 외국어교육에 있어 매우 효과적이다. 둘째, 교육용 로봇 기반 학습시스템은 학습자에게 자기 주도학습의 동기를 부여하였다. 셋째, 사전 테스트와 사후 테스트 결과는 제안된 학습시스템이 로봇과 인공지능 앱을 기반으로 한 학습시스템이 학생들의 성취도 향상에 효과적이었다.

This study reports the effects of text chat on EFL students’ writing fluency, accuracy, and complexity, investigating whether its effects differ according to the interlocutor. The experimental design employed three text chat groups: one between two nonnative speakers (NNS-NNS); another between a nonnative speaker and a native speaker (NNS-NS); and the other between a nonnative speaker and a nativelike chatbot (NNS-NC). 78 college students of English as a foreign language between 19 to 22 years old were sampled and assigned into the three groups, each consisting of 30, 20, and 28, respectively. Over a 16-week period, they engaged in ten 10-minute-long chat sessions. All groups were tested before and after the treatment. A paired-samples t-test was conducted to compare preand post-test scores as far as fluency, accuracy, and complexity concerned. To find out the differences between mean scores of the groups, a one-way analysis of covariance (ANCOVA) was run. Results indicated that all three groups showed significant improvement in accuracy while only NNS-NS and NNS-NC groups did in fluency. No effects for complexity were evident. In terms of group differences, no statistical significance was detected. Participants’ perceptions of English learning and text chat positively changed overall. This study has pedagogical implications for EFL teachers, students, and researchers.

The current study explores the effects of different types of voice-based chat on EFL students’ negotiation of meaning according to proficiency levels. Participants included 123 Korean university students of English. They were divided into two voice-based chat groups: student-student voice-based chat and student-chatterbot voice-based chat. The experiment was administered throughout one semester, 16 weeks. Negotiation of meaning evident in the chats was coded for confirmation check, comprehension check, clarification requests, repetition, and reformulation, and was measured by counting the number of meaning negotiation moves. Important findings were as follows: Firstly, there were significant differences between the first chat and the last chat. The mean frequencies of negotiation moves at all proficiency levels positively changed over time as a result of participating in student-chatterbot voice-based chat. Particulalry, student-chatterbot voice-based chat, as compared to student-student voice-based chat, allowed students to use more negotiation strategies, and the strategies used in the chats also appeared to be different according to the students’ proficiency levels. Lastly, positive perceptions of voice-based chat were observed at all proficiency levels. This study provides empirical evidence to substantiate the effects of voice-based chatterbots in oral interaction. Based on the findings, pedagogical implications are made on the effective implementation of voice-based chatterbots in EFL contexts.

This study investigates how voice-chat conditions affect learners’ affective factors. In order to examine the effects of voice-chat conditions on Korean EFL learners’ attitudes towards English language learning and perceptions of voice-chat, 123 college students participated in this study. Participants were divided into two experimental groups: voice-chat with peers and voice-chat with robots. During the 16-experiment week, participants had a chat with either peers or robots. All participants in both voice-chat conditions showed positive attitudes towards English language learning and positive perceptions of voice-chat. Results indicate that both voice-chat conditions are effective in enhancing the learners’ confidence in English speaking, motivation to develop their English language skills, interests in English language learning, and beliefs in the improvement of their English speaking ability. In addition, engagement in voice-chat appears to help learners to reduce their stress and anxiety level. Findings suggest that EFL teachers integrate chat robots into their language teaching process.

This paper examines the use of English determiners in chat with the focus on determiner-related errors. For this, the comparisons were made in the use of determiners between chat and the grammatical judgment test, which was followed by the investigation of the role of task type on the determiner use in chat. The results show that deletions of determiners are significantly more common in the meaning-focused activity (i.e. chat), and that students commit more errors in the form-focused activity (i.e. the grammatical judgment test). It was also found that task type was a significant factor to cause the difference in the frequency of determiners, implying that the convergent task might be more appropriate to use to look at students’ competence regarding determiner use than the divergent task. However, the overall ratio of deletions was so high that it seems to be hard for researchers to obtain the precise picture of learners’ developmental state through the chat transcript.