Humanlike conversational artificial intelligence agents for foreign or second language learning : user perceptions, expectations, and interactions with agents
Abstract/Contents
- Abstract
- In this dissertation I explore the impact of conversational artificial intelligence (AI) agents on foreign and second language (L2) learning. As products featuring AI continue to gain traction in L2 education, it is vital to understand how the development and adoption of conversational AI has impacted learners and teachers, and how improvements in technical capabilities might affect the future of L2 education. Principles of L2 acquisition developed through decades of theory and empirical studies describe how interaction between learners and other humans drives acquisition of the L2, but it is not yet known how these principles might or might not transfer when learners instead interact with intelligent computer agents. I review literature from second language acquisition (SLA) theory, L2 teaching approaches, computer-assisted language learning, and human-AI interaction, and identify five research questions that focus on knowledge gaps relating to the use of conversational AI for L2 learning. The research questions and the studies they motivate shed light on the nature of L2 learners' interactions with conversational AI agents and how they differ from interactions with human conversation partners, the subjective experience felt by learners interacting with conversational AI agents and human partners, learner expectations and perceptions of conversational AI agents, and the role of conversational AI agents within communicative language teaching and task-based language teaching approaches to L2 education. I describe the results of two studies designed to address the five research questions. In Study 1, I survey 107 experienced L2 learners and 48 experienced L2 teachers about their experiences with, and beliefs about, conversational AI for L2 learning purposes. These language learners and teachers represented a wide range of native languages (L1s) and L2s, and both foreign language and second language contexts. Results indicated that learners and teachers share beliefs about the strengths and weaknesses of both currently available conversational AI products, as well as hypothetical advanced conversational AI products that may be available in the future. Strengths of conversational AI include convenience and availability, high effectiveness for learning, and reduction of anxiety or embarrassment. Weaknesses include low technical capabilities, the lack of human connection and accountability, and the lack of cultural or social aspects. In Study 2, I build a virtual world web application to study how L2 learners' perceptions of their interlocutor as controlled by a human or AI affect the expected pedagogical value of their interactions with the interlocutors. 36 L1 speakers of Chinese who had lived in the United States for at least 6 months and had high-beginner or low-intermediate proficiency in English (L2) completed two language tasks, one each with a perceived human and perceived AI conversation partner. These tasks mimicked important real-life language tasks where the learners would be required to use the L2. I measured the learners' use of conversational strategies and interactional oral feedback, subjective experience, and language quantity and quality during the two language tasks. Results showed that learners used more time-gaining and self-repairing strategies, as well as a higher number of words per conversational turn, when interacting with perceived human partners as compared to interacting with perceived AI partners. These results fit with a pattern established in previous research of users exerting more effort in maintaining conversation with human partners than with computer partners. The overall conclusions of this dissertation are that conversational AI may be effective for developing proficiency in the linguistic aspects of L2 conversational skills, but that important non-linguistic aspects related to fundamental humanness limit the ultimate role of conversational AI in L2 learning. Regardless of how AI technology progresses in the future, conversational AI and humans are likely to both play uniquely valuable roles in the L2 learning process.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2022; ©2022 |
| Publication date | 2022; 2022 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Davis, Glenn Michael |
|---|---|
| Degree supervisor | Padilla, Amado M |
| Thesis advisor | Padilla, Amado M |
| Thesis advisor | Landay, James A, 1967- |
| Thesis advisor | Schwartz, Daniel L |
| Degree committee member | Landay, James A, 1967- |
| Degree committee member | Schwartz, Daniel L |
| Associated with | Stanford University, Graduate School of Education |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Glenn M. Davis. |
|---|---|
| Note | Submitted to the Graduate School of Education. |
| Thesis | Thesis Ph.D. Stanford University 2022. |
| Location | https://purl.stanford.edu/dj743bg5280 |
Access conditions
- Copyright
- © 2022 by Glenn Michael Davis
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...