LeeY2021
| LeeY2021 | |
|---|---|
| BibType | ARTICLE |
| Key | LeeY2021 |
| Author(s) | Yeji Lee |
| Title | Patterns of Users’ First Turns with a Service Chatbot: A Conversation Analytic Perspective |
| Editor(s) | |
| Tag(s) | EMCA, HCI, Chatbots |
| Publisher | |
| Year | 2021 |
| Language | English |
| City | |
| Month | |
| Journal | Korean Journal of Applied Linguistics |
| Volume | 37 |
| Number | |
| Pages | 117-160 |
| URL | Link |
| DOI | 10.17154/kjal.2021.7.37.Special.117 |
| ISBN | |
| Organization | |
| Institution | |
| School | |
| Type | |
| Edition | |
| Series | |
| Howpublished | |
| Book title | |
| Chapter | |
Abstract
The present study examines patterns of users’ first turns in their interaction with a service chatbot developed by a team in the Department of Computer Science and Engineering at Sogang University, South Korea. The user’s first turn is where the user produces their initial request to the chatbot based on a task prompt that is given to them. This is a crucial site that can project the trajectory of the conversational dialogue in the next turns. The data for this study is a corpus of 456 conversational dialogues between human users and the service chatbot for the task of scheduling (e.g., creating/deleting/changing schedules). Analyses reveal three main patterns emerging from users’ first turns: (1) naming main request; (2) prefacing another request; and (3) aggregating task prompt. These patterns are described as users’ sense-making practices which demonstrate their understanding of the task prompt presented to them as well as how they interpret the underlying mechanism of the chatbot. The first and second patterns, in particular, are illustrative of discrepancies in assumptions between human users and chatbot developers. The study provides practical implications for chatbot developers and discusses the utility of Conversation Analysis (CA) as a methodology to investigate human-chatbot interaction.
Notes
