Abstract:

Introducing dialogue-based computer-assisted language learning enables meaning-focused speaking practice in foreign language education. It is vital to understand how students experience this educational activity. Spoken dialogue systems (SDSs), targeting natural language interfaces for human-machine interaction, provide opportunities for practising speaking [1]. Students interact with human-like interlocutors known as embodied conversational agents (CAs) to solve tasks in everyday-life situations, such as ordering food in a restaurant. This aligns with ideas of task-based language teaching, focusing on implicit knowledge for the ability to speak and gain fluency [2]. Previous research, mostly conducted in higher education, has recognised the effectiveness of SDSs [3]. However, little known empirical research has focused on exploring students’ educational experiences. This research aims to provide a nuanced and comprehensive understanding of students' experiences practising foreign language speaking skills in an SDS. The questions guiding this research are:

(i) How do lower secondary school students experience practising speaking skills in an SDS?
(ii) How do lower secondary school students experience speaking with a CA?

Three studies were conducted with two different SDSs in educational settings with Swedish lower secondary school students (N=88) self-reporting their experiences. Data were produced through digital logbooks, questionnaires, and system-generated metrics, complemented with interviews. Qualitative data were analysed through thematic analysis, and quantitative data through descriptive statistics. In the third study, an analytical lens was applied where the students’ longitudinal experiences were dimensionalised into cognitive, emotional, social, and teaching experiences, altogether forming the educational experience. Data were analysed through inferential statistics.

The synthesised results generated a model illustrating interrelated central aspects of the students’ educational experiences, organised into three areas
(i) System functionality,
(ii) Learning, and
(iii) Engagement. There were within-individual and between-individual differences in the identified aspects of the three areas. This research implies that the pedagogical and individual framing of this educational activity is important, although SDSs have been shown to facilitate self-regulated learning. One practical implication of this model is for teachers and stakeholders to gain an understanding of students’ educational experiences of practising speaking in an SDS. The findings can be viewed as inspiration for dialogue-based computer-assisted language learning and be applied when integrating SDSs and designing them. Further studies are encouraged to test and refine this suggested model.

References:
[1] M. McTear, M. Conversational AI: Dialogue systems, conversational agents, and chatbots. Morgan & Claypool Publishers. 2021.
[2] R. Ellis, P. Skehan, S. Li, N, Shintani & C. Lambert. Task-Based Language Teaching: Theory and Practice. Cambridge University Press.2021.
[3] S. Bibauw, T. François, W, Van den Noortgate & P. Desmet. Dialogue systems for language learning: a meta-analysis. Language Learning & Technology, 26(1), 1-24, https://hdl.handle.net/10125/73488

Keywords:

Conversational agents, educational experience, model, speaking skills, spoken dialogue systems, students.