See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/386135208

Examining the effect of generative AI on students' motivation and writing

self-efficacy

Article · December 2024

DOI: 10.29140/dal.v1.102324

CITATIONS READS

0 49

2 authors:

Jerry Yung Teh Huang Atsushi Mizumoto

Kansai University Kansai University
6 PUBLICATIONS 6 CITATIONS 50 PUBLICATIONS 1,112 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Jerry Yung Teh Huang on 26 November 2024.

The user has requested enhancement of the downloaded file.

 Digital Applied Linguistics
ISSN 2982-2300

https://www.castledown.com/journals/dal/

Digital Applied Linguistics, 1, 102324 (2024)

https://doi.org/10.29140/dal.v1.102324

Examining the effect of generative AI on students’ motivation

and writing self-efficacy
JERRY HUANGa (Corresponding Author)
ATSUSHI MIZUMOTOa
a
Kansai University, Faculty of Foreign Language Studies, Japan

Abstract
The present study explores the effects of generative AI, specifically ChatGPT, in EFL
classrooms on student motivation and writing efficacy. Motivation was measured
through three components: Ideal L2 Self (IL2), Ought-to L2 Self (OL2), and L2 Learning
Experience (L2LE). Participants (n = 327) were first and second-year undergraduate
students at a Japanese university, enrolled in mandatory English classes focused on
reading/writing or speaking/listening. The control group (n = 164) received peer
feedback, whereas the treatment group (n = 163) utilized ChatGPT with specially crafted
prompts for feedback. Both groups completed pre- and post-questionnaires to assess
motivation and writing self-efficacy. Results affirmed that ChatGPT positively affected
students’ motivation related to Ideal L2 Self and L2 Learning Experience. ChatGPT also
significantly enhanced writing self-efficacy, which was found to correlate with all three
motivational factors. However, there was no impact on Ought-to L2 Self motivation.
The study highlights that ChatGPT’s integration can improve intrinsic motivation and
writing self-efficacy, provided structured guidance is available to manage issues such
as plagiarism. Future research should examine diverse samples, long-term effects, and
ChatGPT’s impact on other language skills.
Keywords: The L2 motivational self system; ideal L2 self; ought-to-L2 self; L2 learning
experience; generative AI; ChatGPT; self-efficacy

Introduction
In the field of digital applied linguistics, the rapid advancement of Artificial Intelligence (AI)
technology has resulted in significant transformations, creating a growing demand for the
integration of AI into writing curriculum development (Hao et al., 2024; Yang et al., 2024).
AI technologies can provide personalized feedback, facilitate collaborative writing experiences,

Copyright © 2024 Received XXX XX, XXXX

Jerry Huang & Atsushi Mizumoto Accepted XXX XX, XXXX
2 Examining the effect of generative AI on students’ motivation and writing self-efficacy

and offer engaging writing prompts, all of which may foster a more motivating learning
environment. Motivation has long been recognized as a critical factor in applied linguistics,
influencing learners’ engagement, persistence, and success (Dörnyei, 2020). With the advent
of digital tools and platforms, it becomes increasingly essential to explore how AI technology
can enhance or hinder motivation among language learners. One of the latest technological
advancements in AI technology is the introduction of AI chatbots, such as ChatGPT, released
in late 2020. These chatbots offer interactive and personalized language practice opportunities,
which can potentially boost motivation by providing immediate feedback, simulating real-life
conversations, and accommodating individual learning paces. Despite the promising potential
of AI chatbots, the specific role of ChatGPT feedback in learners’ motivation remains largely
under-researched.
A recent study conducted in the United States (McDonald et al., 2024) surveyed 116 top
research (R1) universities. Findings show that only a modest number (63%) actively encourage
the use of generative AI (GenAI), and roughly half offer extensive classroom guidance, sample
syllabi, and curriculum support, primarily centered on writing activities. However, the
widespread availability of smartphones allows learners easy access to information and tools,
such as computer-assisted language learning, both inside and outside the classroom. Common
resources such as Google Translate and DeepL are already widely used among learners, and
studies (e.g., Stapleton & Leung, 2019) have thoroughly examined their effectiveness. With
the rise of GenAI, learners now have access to a broader range of tools that can enhance their
language learning experience. Recent research indicates that ChatGPT may foster learner
autonomy (Augustini, 2023), enhance students’ motivation and writing skills (Song & Song,
2023), and encourage pleasure-based motivation (Cai et al., 2023).
Although the use of ChatGPT appears to positively influence motivation, the commonly
referenced L2 Motivational Self-System (L2MSS) in second language acquisition (SLA) has
rarely been applied in this context. Huang and Mizumoto (2024a) demonstrated ChatGPT’s
effectiveness in maintaining student motivation. The present study builds upon their work by
expanding the participant pool and adding an analysis of learners’ perceptions of their self-
efficacy in writing after using ChatGPT.
This study seeks to provide a more comprehensive understanding of how ChatGPT influences
learners’ motivation. By expanding the participant pool and incorporating an analysis of
learners’ perceptions of their self-efficacy in writing after using ChatGPT, it offers a deeper
examination of the alignment between ChatGPT’s impact and the principles of the L2MSS.
Additionally, the study explores how learners’ motivational self-systems interact with AI-driven
language learning environments.

Literature Review
L2 Motivational Self System
Following Zoltán Dörnyei (2005), the L2 Motivational Self System (L2MSS) has become a major
focus of study, spurring substantial research on individual differences in language learning
motivation (Al-Hoorie, 2018). Although questions have been raised about the instrument’s
validity (Al-Hoorie et al., 2023), other researchers have reaffirmed the system’s reliability
(Henry & Liu, 2024; Papi & Teimouri, 2024). Despite recent debates, this framework remains
a significant development in SLA as it explores the psychological factors that drive learners’
motivation to study a second language (L2). Comprising three components—the Ideal L2 Self,
the Ought-to L2 Self, and the L2 Learning Experience—the L2MSS examines how different
aspects of learners’ self-perception influence motivation. The Ideal L2 Self, seen as the
strongest motivating factor, represents learners’ envisioned future self as a fluent language

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 3

user. The Ought-to L2 Self reflects external pressures and expectations about language
proficiency, while the L2 Learning Experience encompasses learners’ real interactions in the
language, which can affect motivation either positively or negatively.
Dörnyei (2005) describes the Ideal L2 Self as the specific part of a learner’s ideal identity tied
to their proficiency in an L2, embodying the inspiring vision of becoming fluent and socially
capable in the language. This aspirational self-image strongly motivates individuals to narrow
the gap between their current skills (actual self) and the ideal self they aim to achieve. Studies
by Taguchi et al. (2009) and Ryan (2009) have found that this concept is closely linked to inte-
grativeness and accounts for a significant portion of the variance in learners’ effort.
The Ought-to L2 Self reflects learners’ sense of duty or obligation to learn an L2, shaped
primarily by external expectations. It represents qualities learners believe they should develop
based on responsibilities or others’ expectations. Taguchi et al. (2009) found in a comparative
study across Japan, China, and Iran that family influence and preventive motivations were key
factors in shaping this aspect. Further research in Hungary by Csizér and Kormos (2009) noted
a positive link between parental support and the Ought-to L2 Self.
The L2 Learning Experience refers to how learners feel about language learning, often influ-
enced by immediate factors such as the classroom environment. According to Dörnyei (2009),
many learners find motivation in successful experiences within the learning process itself rather
than through pre-existing or externally imposed self-images. This aspect is affected by factors
like the curriculum, the L2 instructor, classmates, and teaching materials.

Generative AI and Writing

The launch of OpenAI’s ChatGPT in November 2022 has revolutionized academic writing,
showcasing its ability to produce “high-quality” papers suitable for journal submissions. Teng
(2023) noted that with ongoing AI advancements, significant changes in academic publishing
are anticipated, suggesting that the field is still in the early stages of transformation. However,
while large language models present exciting potential for enhancing education and support-
ing instructors, their limitations and potential biases must be carefully considered to ensure
responsible usage (Kasneci et al., 2023). Yang et al. (2024) studied the integration of ChatGPT
in an EFL academic writing course, finding it beneficial for tasks like proofreading, brainstorm-
ing, and translation. However, the study stresses the importance of using ChatGPT thought-
fully, framing it as a “ghostwriter” that should be navigated with caution. Prioritizing learners’
development as critical thinkers and emerging scholars remains essential to avoid over-reliance
on AI and to foster genuine academic growth.
This new technology has generally been met with enthusiasm. Tseng and Warschauer (2023)
introduced a five-part educational framework to help L2 learners engage with AI by focusing
on understanding, accessing, prompting, verifying, and integrating AI effectively. By teaching
students how to interact with AI, educators can prepare them for technology’s evolving role both
in and out of the classroom. Additionally, Cotton et al. (2023) underscores the need for clear
communication about assessment standards and academic integrity to minimize issues such
as cheating in university settings. As these pedagogical challenges emerge, Hao et al. (2024)
advocate for a balanced approach to AI integration within English major programs, emphasizing
the importance of developing both AI proficiency and critical thinking skills to prepare students
for a technology-driven workforce. The authors propose a theoretical and pedagogical model
that supports AI and critical thinking integration through a dynamic learning environment,
curriculum evolution, research enhancement, and faculty and student support.
Research on ChatGPT is in its early phases but reveals significant potential for educational
development, particularly in L2 learning. ChatGPT has been shown to enhance learner

Digital Applied Linguistics, Volume 1 (2024)

4 Examining the effect of generative AI on students’ motivation and writing self-efficacy

perceptions (Marzuki et al., 2023; Xiao & Zhi, 2023; Van Horn, 2024), boost writing and
language skills (Athanassopoulos et al., 2023; Song & Song, 2023), and support teachers in L2
instruction (e.g., Barret et al., 2023; Jeon & Lee, 2023; Mohamed, 2024). ChatGPT’s integration
into the writing process, from planning to editing, offers structured support while maintaining
integrity (Barrot, 2023). Huang (2023) demonstrated effective pedagogical practice by using
well-designed prompts to help students receive personalized feedback on their work. These
findings underscore ChatGPT’s growing role as a valuable tool for enhancing both teaching and
learning experiences in L2 education.
As more studies emerge, the technology’s role in education is likely to expand, setting new
standards for instructional methods. For example, Kohnke et al. (2023) found ChatGPT use-
ful for enhancing students’ comprehension and assisting teachers with lesson planning. Guo
and Wang (2023) suggest that teachers incorporate ChatGPT’s essay feedback into their own
responses to students. This technology is positioned to make a lasting impact, prompting ques-
tions about its effects on L2 learner motivation and identity.
There were some studies specifically on ChatGPT and L2 writing. Research into the use of
ChatGPT for writing assessment has shown promising results. Teng (2024b) explores the role
of ChatGPT in EFL writing through a systematic review of the 20 most relevant articles, fol-
lowing PRISMA guidelines. The findings indicated that while ChatGPT provides opportunities
for enhancing writing skills—such as instant feedback and diverse prompts—it also presents
challenges, including potential dependency on AI and the necessity for critical thinking, thus
advocating for a balanced approach to integrating AI tools like ChatGPT into writing curric-
ula and fostering a community of practice among educators and students as recommendation.
Mizumoto and Eguchi (2023) investigated its application in L2 writing environments, specif-
ically focusing on Automated Essay Scoring (AES). By analyzing a dataset of 12,100 essays,
they found that AES systems powered by GPT technology demonstrated significant accuracy
and reliability, closely matching human evaluations. This suggests that ChatGPT can provide
automated corrective feedback (CF), making it a valuable resource for writing instructors in
educational settings. Building on this, Mizumoto et al. (2024) assessed ChatGPT’s effectiveness
in evaluating accuracy in L2 writing. Using the Cambridge Learner Corpus First Certificate
in English (CLC FCE) dataset, they compared ChatGPT’s performance with human evaluators
and Grammarly. The study revealed a high correlation between ChatGPT’s assessments and
human ratings, with a correlation coefficient (ρ) of 0.79, surpassing Grammarly’s 0.69. This
highlights ChatGPT’s precision in automated evaluations, confirming its suitability as a tool
for L2 writing assessment. In another study, Allen and Mizumoto (2024) examined the experi-
ences of 33 Japanese EFL learners who used writing groups alongside AI technology for editing
and proofreading. The students expressed a preference for AI tools, like ChatGPT, for editing
and proofreading, appreciating the effective feedback that improved the clarity and cohesion
of their writing. ChatGPT provided authoritative insights and feedback, similar to peer review
processes in writing groups. The authors recommended integrating ChatGPT with writing
groups to enhance editing and proofreading practices. Additionally, Teng (2024c) explored the
impact of ChatGPT on 45 EFL learners in Macau, focusing on their perceptions and experiences
with AI-generated feedback. The findings highlighted significant positive effects of AI assis-
tance on writing, including increased motivation, self-efficacy, engagement, and a tendency
towards collaborative writing. Qualitative analysis identified four key themes: the influence of
AI assistance on writing self-efficacy, engagement, motivation, and the potential for collabora-
tive writing. These studies collectively underscore the transformative potential of ChatGPT in
enhancing writing skills and educational practices.

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 5

Generative AI Use and Motivation

The use of GenAI in EFL classrooms is still emerging, but research has already shown a
positive connection between AI and student motivation. For instance, a study by Leong
et al. (2024) investigated whether GenAI could personalize vocabulary-learning examples
to enhance both learning outcomes and motivation. Researchers developed a web app
with three conditions: a control group using pre-existing example sentences, and two
experimental groups where participants generated sentences or short stories based on their
interests. Although no significant difference in learning outcomes was found across the
conditions, participants in the AI-supported groups reported higher intrinsic motivation, a
stronger sense of choice, and greater feelings of competence compared to the control group.
This suggests that while AI personalization may not directly affect learning outcomes, it
can positively influence students’ motivation. Another study by Wei (2023), conducted at
a university in China, found that AI-mediated language instruction using the Duolingo
platform significantly improved English learning achievement, L2 motivation, and self-
regulated learning among EFL learners, compared to traditional methods. This was further
supported by qualitative interviews with students in the experimental group, who viewed the
AI platform as engaging, personalized, and empowering, which led to increased motivation,
confidence, and a more positive learning environment. Additionally, Yamaoka (2024) found
that ChatGPT acted as a personalized teaching assistant, boosting Japanese EFL learners’
motivation and reducing their anxiety by allowing them to take control of their learning.
However, some learners experienced negative motivational impacts, such as anxiety over
potential misinformation or a loss of confidence due to ChatGPT’s high-quality proofreading.
Together, these studies highlight the positive impact of GenAI on learning motivation,
despite potential challenges.
The studies above focus on the influence of AI on student motivation, rather than the reverse.
A study by Zheng et al. (2024) investigated the technology acceptance of ChatGPT among
EFL learners, specifically examining the factors outlined in the Unified Theory of Acceptance
and Use of Technology 2 (UTAUT2) model, as well as the moderating role of motivation as
conceptualized by Self-Determination Theory (SDT). The results showed that performance
expectancy, effort expectancy, social influence, hedonic motivation, habit, and SDT motivation
were significant predictors of learners’ behavioral intention to use GenAI tools. Additionally,
the study found that SDT motivation significantly moderated the relationships between
certain UTAUT2 constructs and learners’ behavioral intention and actual use of GenAI
tools, highlighting the important role of motivation in shaping technology acceptance. Lai
et al. (2023) explored the factors influencing Hong Kong undergraduate students’ intentions
(as defined by behavioral intent in the Technology Acceptance Model) to use ChatGPT to
support active learning. The study found that intrinsic motivation and perceived usefulness
were the strongest predictors of students’ intention to use ChatGPT for answering academic
inquiries. Similarly, Huang and Mizumoto (2024b) found that in Japanese EFL classrooms
where ChatGPT is a required learning tool, Ought-to L2 Self motivation became a significant
predictor of technology acceptance, specifically influencing students’ actual use of ChatGPT.
This suggests that when students feel a sense of obligation or external pressure to learn
English, they are more likely to adopt new technologies to meet these expectations. However,
the study emphasized the need to explore additional components within the L2MSS that
might yield similar effects. These studies further reinforce the connection between GenAI
and motivation.

Digital Applied Linguistics, Volume 1 (2024)

6 Examining the effect of generative AI on students’ motivation and writing self-efficacy

Feedback and Motivation

Feedback and motivation are closely related. Fang (2023) examined how written feedback impacts
writing motivation and self-efficacy among Chinese EFL high school students. The findings showed
that while feedback significantly increased students’ motivation to write, it did not substantially
influence their self-efficacy in writing. This suggests that feedback can encourage students to
engage more in writing but may not necessarily improve their confidence in their writing abilities.
Similarly, Ahmetovic et al. (2023) explored the role of CF in motivation and EFL achievement
among middle and high school students in Bosnia and Herzegovina. The study revealed that stu-
dents generally appreciate CF and view it positively, believing it enhances their learning. While a
positive attitude toward CF was linked to increased motivation for speaking and writing in English,
neither oral nor written CF significantly predicted students’ overall EFL achievement. A meta-anal-
ysis by Cen and Zheng (2023) offers further insight into the relationship between feedback and
motivation. The study highlighted several important points: when delivered effectively, feedback
can be a powerful motivator; personalized feedback is essential, as its impact varies among stu-
dents; students who view feedback positively are generally more motivated; feedback encourages
self-regulation, which indirectly boosts motivation; and a supportive classroom environment can
enhance the motivational benefits of feedback. These findings suggest that GenAI could contribute
meaningfully by providing personalized feedback, reducing anxiety, promoting positive attitudes,
and fostering a supportive context. Thus, feedback from GenAI could indirectly enhance students’
motivation, aligning with the patterns observed in previous studies.

Motivation and Writing Self-Efficacy

Self-efficacy, a central concept in social cognitive theory, refers to individuals’ beliefs in their
ability to organize and execute actions to reach specific goals (Bandura, 1997). These beliefs
have a strong influence on performance, often predicting outcomes more reliably than actual
skill level (Bandura, 1997; Schunk, 1991). Research indicates that self-efficacy is not only a cru-
cial motivational factor but also shapes students’ choices, effort, persistence, thought processes,
and emotional responses to writing tasks (Zhang et al., 2023). In an investigation of online
English learning among Chinese university students, Teng (2024a) explored the relationships
between social support, self-efficacy, emotional adjustment, and foreign language anxiety
(FLA). The findings show that social support significantly predicts emotional adjustment and
self-efficacy beliefs, and that self-efficacy acts as a mediator between social support and FLA.
This suggests that fostering self-efficacy through social support may be an effective way to alle-
viate FLA and improve outcomes in online English learning. Additionally, a study by Teng and
Yang (2022) examined the interplay among metacognition, motivation, self-efficacy beliefs, and
English learning outcomes in online learning environments. Their results revealed that self-ef-
ficacy significantly predicts online English learning achievement, both directly and indirectly,
through motivation and metacognition. This study underscores the value of a supportive online
learning environment that cultivates self-efficacy, motivation, and metacognitive awareness to
enhance learners’ experiences. Given these insights, the introduction of GenAI in EFL settings
may also contribute positively. As Huang and Mizumoto (2024a) highlight, increased motiva-
tion could, in turn, lead to enhanced self-efficacy in writing, showing the potential for GenAI to
support and elevate learning in EFL contexts.

Research Questions
Previous research showed a slight increase in all three motivation factors, though the increase
in the Ought-to L2 self was not statistically significant. This study aims to replicate those find-
ings while also comparing students’ self-reported writing efficacy before and after using GenAI.

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 7

Additionally, it also seeks to investigate the relationship between motivation and self-efficacy
following GenAI usage. Based on the literature review, the research questions are as follows:

RQ1: Do students report higher Ideal L2 Self motivation after using GenAI in the
language classroom?
RQ2: Do students report higher Ought-to L2 motivation after using GenAI in
the language classroom?
RQ3: Do students report higher L2 Learning Experience motivation after
using GenAI in the language classroom?
RQ4: Do students report higher writing self-efficacy after using GenAI in the
language classroom?
RQ5: D o the correlations between writing self-efficacy and motivational fac-
tors increase after using GenAI in the language classroom?

Materials and Methods

Participants
This study involved 327 first- and second-year undergraduate students taking a required
English course. Of these, 164 students in a Listening/Speaking class formed the control group,
while 163 students in a Reading/Writing class made up the treatment group. In the Listening/
Speaking class, students individually prepared scripts for group presentations and received
feedback from their peers. Meanwhile, in the Reading/Writing class, students collaborated
on essays, with each student responsible for a different paragraph, such as the introduction,
body, or conclusion. Both groups completed their respective tasks twice over the course of the
semester.

Instruments and Data Collection

Students completed a survey twice during the semester—once at the beginning and once at
the end—to capture potential changes in motivation and writing self-efficacy over time. The
study used a 22-item questionnaire (see Appendix), adapted from the works of Taguchi et al.
(2009) and Pajares and Valiante (1999). Taguchi et al.’s (2009) L2 Motivational Self System
(L2MSS) questionnaire, initially developed for Japanese participants, provided the framework
for assessing three core motivational components: Ideal L2 Self (IL2), Ought-to L2 Self (OL2),
and L2 Learning Experience (L2LE).

• IL2 items prompted students to envision themselves confidently using English

in their future, with questions like, “I can imagine myself living abroad and
having a discussion in English.”
• OL2 items reflected external pressures or expectations, such as “I study English
because close friends of mine think it is important.”
• L2LE items focused on students’ current engagement in English classes, with
items like “I really enjoy learning English.”

In addition, ten questions were adapted from Pajares and Valiante’s (1999) writing self-efficacy
scale to assess students’ confidence in specific writing skills, such as “I can structure paragraphs
to support ideas in the topic sentences” and “I can write simple sentences with good grammar.”
To ensure the integrity of responses and minimize potential biases, the university’s online
learning management system randomized the sequence of questions for each participant. Any

Digital Applied Linguistics, Volume 1 (2024)

8 Examining the effect of generative AI on students’ motivation and writing self-efficacy

incomplete survey prompted a reminder for students to finish. All survey questions were pre-
sented in Japanese, and responses were recorded on a 6-point Likert scale, with options rang-
ing from 1 (“strongly disagree”) to 6 (“strongly agree”).

Comparison of Control and Treatment Groups

Participants in this study were students from a private university in Japan’s Kansai region,
enrolled in either a required English reading and writing class or a speaking and listening class.
While individual English proficiency was not directly measured, it was estimated to range from
A2 to B2 on the CEFR scale, based on guidelines from Aizawa et al. (2020). All participants had
completed at least eight years of English instruction as part of Japan’s mandatory education
curriculum in primary and secondary schools.
Students in the control group participated in speaking and listening classes, which included
two group oral presentations during the semester. Presentation topics were aligned with the
university’s prescribed textbook to maintain consistency. Prior to each presentation, students
used an entire class period to develop their speaking scripts collaboratively, receiving peer feed-
back in groups of four or five. This feedback was intended to improve transitions and script
coherence, moving beyond basic comments. Based on their peers’ input, students revised their
scripts, completing them either in class or as homework. All tasks were done on paper, and final
scripts were submitted as part of the assessment after the oral presentations.
The treatment group consisted of students in reading and writing classes, where they col-
laborated on two group essays over the semester. Essay topics were similarly aligned with the
textbook to support writing skills development. Each group of five students worked on one part
of the essay, with each student responsible for a specific paragraph (e.g., introduction, body,
or conclusion). On the first day of the writing process, students created a flowchart, outlined,
and drafted their essays on paper to limit reliance on ChatGPT. Any incomplete work could be
finished at home. On the second day, students attended a brief lecture on using ChatGPT, with
pre-made prompts focused on providing structured feedback on essay organization, coherence,
and structure (Huang, 2023). Students received training on plagiarism prevention and were
required to avoid copy-pasting text. Any revisions based on AI feedback were documented on
paper and submitted through the university’s online learning system for further review. By the
end of the second week, students submitted the final version of their essays along with paper
drafts showing their revision history.

Data Analysis
The data for this study was processed using R software (version 4.3.3). To answer the first
four research questions, the dataset was analyzed for normality through Shapiro-Wilk p-val-
ues (w) and for homogeneity of variance using standard deviations. Instrument reliability
was assessed by calculating Cronbach’s alpha (α). Interaction effects between factors were
explored using a two-way analysis of variance (ANOVA), which was appropriate due to the
dataset meeting the assumptions of normality, homogeneity, and reliability. This approach
enabled the comparison of mean scores between two groups and assessed the statistical sig-
nificance of post-intervention differences. To address the fifth research question, Pearson’s
correlation coefficient (r) was employed to determine the nature and strength of the relation-
ship between motivation and self-efficacy in both groups before and after the intervention.
This analysis was essential for evaluating how the intervention influenced these psychological
constructs.
To ensure the reproducibility and transparency of the data analysis process, the data and the
R code used in the study have been made accessible on OSF (https://osf.io/h4up2).

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 9

Results
The following descriptive statistics provide insights into the effects of treatment and time on
three L2MSS constructs and writing self-efficacy. Table 1 shows that the mean scores for these
variables range from 3.36 to 4.10, with standard deviations between 0.81 and 1.34, indicating
relatively consistent variance across the groups. Shapiro-Wilk test p-values, ranging from
0.97 to 0.99, suggest a normal distribution for all variables within each group. Additionally,
Cronbach’s alpha (α) values, which range from 0.75 to 0.92, demonstrate strong internal
consistency and reliability of the variables measured. The treatment group’s mean scores show
notable increases across all four variables (3.55 to 3.71, 3.42 to 3.61, 4.29 to 4.47, and 3.52 to
3.95), whereas the control group’s mean scores remained relatively stable with slight variability
(3.55 to 3.55, 3.36 to 3.43, 4.24 to 4.22, and 3.63 to 4.10). Given the normal distribution,
homogeneity, and reliability of the data, ANOVA can be used to test for significant differences
and interaction effects, assessing whether the post-intervention score increases in these
variables are statistically significant.
A two-way ANOVA was performed to examine the interaction between ChatGPT usage and
pre-post scores for each variable. The findings are presented in Table 2.
The results of Ideal L2 self in Table 2 shows no significant effect of treatment alone (p = .49)
with a negligible effect size (Partial η² < .01). However, time had a significant effect (p < .05)
with a small effect size (Partial η² = .01), indicating that changes over time are meaningful.

Table 1 Descriptive statistics.

Variables Group Test n Mean SD p α

Ideal L2 Self Treatment Pre 163 3.55 1.03 Pre .98 Pre .86
Post 163 3.71 1.13
Control Pre 164 3.55 1.08 Post .98 Post .88
Post 164 3.55 1.14
Ought-to L2 Treatment Pre 163 3.42 1.16 Pre .98 Pre .75
Self
Post 163 3.61 1.3
Control Pre 164 3.36 1.13 Post .97 Post .85
Post 164 3.43 1.34
L2 Learning Treatment Pre 163 4.29 0.96 Pre .98 Pre .88
Experience
Post 163 4.47 1.09
Control Pre 164 4.24 1 Post .97 Post .89
Post 164 4.22 1.02
Writing Self- Treatment Pre 163 3.52 0.86 Pre .99 Pre .91
Efficacy
Post 163 3.95 0.93
Control Pre 164 3.63 0.81 Post .99 Post .92
Post 164 4.10 0.84

Digital Applied Linguistics, Volume 1 (2024)

10 Examining the effect of generative AI on students’ motivation and writing self-efficacy

Table 2 Two-way ANOVA with interaction (Ideal L2 self).

Source SS df MS F-ratio p-value Partial η2

Treatment 1.02 1 1.02 0.48 .49 < .01
Time 1.12 1 1.12 3.96 <.05 .01
Treatment × Time 1.06 1 1.06 3.73 .05 .01

Table 3 Post analysis (Ideal L2 self).

Source SS df MS F-ratio p-value Partial η2

Treatment at Pre < 0.01 1 < 0.01 < 0.01 .99 < .01
Treatment at Post 2.07 1 2.07 1.61 .21 < .01
Time in Control < 0.01 1 < 0.01 < 0.01 .97 < .01
Time in Treatment 2.17 1 2.17 7.17 .01 .04

The interaction between treatment and time approached significance (p = .05) with a small
effect size (Partial η² = .01). Despite the marginal significance of this interaction, post-hoc tests
were conducted to explore potential differences further. This approach aimed to examine subtle
interactions and offer a more thorough understanding of the data, as near-significant findings
may reveal relevant trends and insights.
Table 3 provides the post-hoc analysis results, highlighting a significant interaction between
the two factors. The analysis indicates that treatment alone had no significant effect at either
the pre-test (p = .99) or post-test (p = .21) stages, with minimal effect sizes (Partial η² < .01).
Time also showed no significant effect within the control group (p = .97), with no measurable
effect size (Partial η² < .01). However, in the treatment group, time had a significant effect
(p = .01) with a moderate effect size (Partial η² = .04), suggesting that ChatGPT usage produced
a meaningful impact on students’ Ideal L2 Self over time.
According to the results in Table 4, treatment alone did not have a significant effect (p =
.30) and showed a negligible effect size (Partial η² < .01). Time, however, showed a significant
effect (p = .04) with a small effect size (Partial η² = .01), indicating that changes over time were
meaningful for both groups. The interaction between treatment and time was not significant
(p = .33) with no notable effect size (Partial η² < .01), suggesting that ChatGPT use had no
impact on students’ Ought-to L2 Self.
The results of L2 Learning Experience in Table 5 indicate that treatment alone had no
significant effect (p = .15) with a very small effect size (Partial η² < .01). Time showed a marginally
significant effect (p = .10) with a similarly small effect size (Partial η² < .01), suggesting slight
changes over time. The interaction between treatment and time, however, was significant
(p < .05) with a small effect size (Partial η² = .01), pointing to a meaningful interaction effect.
Consequently, a post-hoc test was conducted, with the results displayed in Table 6.
The post-hoc test results indicate that treatment had no significant effect at the pre-test stage
(p = .64, Partial η² < .01). However, at the post-test stage, treatment showed a significant effect

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 11

Table 4 Two-way ANOVA with interaction (Ought-to L2 self).

Source SS df MS F-ratio p-value Partial η2

Treatment 2.65 1 2.65 1.09 .30 < .01
Time 2.74 1 2.74 4.46 .04 .01
Treatment × Time 0.57 1 0.57 0.94 .33 < .01

Table 5 Two-way ANOVA with interaction (L2 Learning Experience).

Source SS df MS F-ratio p-value Partial η2

Treatment 3.58 1 3.58 2.13 .15 < .01
Time 1.10 1 1.10 2.78 .10 < .01
Treatment × Time 1.55 1 1.55 3.91 <.05 .01

Table 6 Post analysis (L2 Learning Experience).

Source SS df MS F-ratio p-value Partial η2

Treatment at Pre 0.21 1 0.21 0.22 .64 < .01
Treatment at Post 4.92 1 4.92 4.41 .04 .01
Time in Control 0.02 1 0.02 0.04 .83 < .01
Time in Treatment 2.62 1 2.62 7.33 <.01 .04

(p = .04, Partial η² = .01). Time had no significant effect in the control group (p = .83, Partial
η² < .01) but showed a significant effect in the treatment group (p < .01, Partial η² = .04).
These findings suggest that using ChatGPT positively impacted students’ L2 Learning Experi-
ence over time.
Table 8 presents the results for Writing Self-Efficacy, showing that treatment alone had no sig-
nificant effect (p = .15) with a negligible effect size (Partial η² < .01). The time factor showed a
marginal effect (p = .10) with a very small effect size (Partial η² < .01). The interaction between
treatment and time was significant (p < .05) with a small effect size (Partial η² = .01), indicating
a meaningful interaction effect between the two factors, which warranted further examination
through a post-hoc test.
The post-hoc test results in Table 9 indicate that treatment had no significant effect at the pre-
test stage (p = .64) with a negligible effect size (Partial η² < .01). At the post-test stage, however,
treatment showed a significant effect (p = .04) with a small effect size (Partial η² = .01). Time
had no significant effect in the control group (p = .83) with a negligible effect size (Partial η²
< .01), whereas time in the treatment group showed a significant effect (p < .01) with a moder-
ate effect size (Partial η² = .04). These results suggest that ChatGPT use was effective in enhanc-
ing students’ writing self-efficacy.

Digital Applied Linguistics, Volume 1 (2024)

12 Examining the effect of generative AI on students’ motivation and writing self-efficacy

Table 7 Two-way ANOVA with interaction (Writing Self-Efficacy).

Source SS df MS F-ratio p-value Partial η2

Treatment 3.58 1 3.58 2.13 .15 < .01
Time 1.10 1 1.10 2.78 .10 < .01
Treatment × Time 1.55 1 1.55 3.91 <.05 .01

Table 8 Post analysis (Writing Self-Efficacy).

Source SS df MS F-ratio p-value Partial η2

Treatment at Pre 0.21 1 0.21 0.22 .64 < .01
Treatment at Post 4.92 1 4.92 4.41 .04 .01
Time in Control 0.02 1 0.02 0.04 .83 < .01
Time in Treatment 2.62 1 2.62 7.33 <.01 .04

Table 9 Correlation of Writing Self-Efficacy with motivational factors.

Group WSE IL2 OL2 L2LE

Pre .47 .23 .47
Control
Post .44 .25 .40
Pre .48 .19 .53
Treatment
Post .50 .35 .60
Note. All values are p < .05

Table 9 presents the Pearson correlation coefficients (r) for Writing Self-Efficacy (WSE) in
relation to three motivational factors—Ideal L2 Self (IL2), Ought-to L2 Self (OL2), and L2
Learning Experience (L2LE)—for both the control and treatment groups, measured before and
after the intervention. In the control group, WSE shows moderate correlations with IL2 (r =
.47) and L2LE (r = .47) at the pre-test stage, which slightly decrease at the post-test stage (r
= .44 for IL2 and r = .40 for L2LE), while the correlation with OL2 slightly increases from r =
.23 to r = .25. In the treatment group, WSE shows moderate correlations with IL2 (r = .48) and
L2LE (r = .53) at pre-test, which both increase at post-test. Notably, the correlation with OL2
rises from r = .19 to r = .35, and with L2LE from r = .53 to r = .60. These results suggest that the
treatment positively strengthened the correlations between WSE and the motivational factors,
whereas the control group exhibited either stable or slightly reduced correlations over time.

Discussion
In addressing RQ1, the treatment group reported a higher mean score, whereas the control
group reported no change in mean score. Although the interaction of the two factors showed

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 13

marginal significance, the post-hoc test identified one significant finding. This suggests that
instructor-led ChatGPT use in an EFL writing class positively influenced students’ Ideal L2 Self
motivation, as students in the treatment group reported higher scores. Hence the answer to this
RQ is “Yes, students do report higher Ideal L2 Self motivation after instructor-led GenAI usage
in the classroom”.
The Ideal L2 Self concept serves as a motivational driver by highlighting the gap between one’s
current state and their desired future self (Dörnyei, 2009). In this study, the treatment group’s
higher mean score indicates that students view this innovative tool as a valuable resource for
academic support. This positive reinforcement helps enhance their self-perception. The tool’s
ease of use allows students to integrate it seamlessly into their studies, increasing their moti-
vation and aligning with the Ideal L2 Self (Taguchi et al., 2009; Ryan, 2009). Conversely, the
control group, which did not use ChatGPT, showed no change in their Ideal L2 Self scores. This
may be due to a lack of engagement, as they experienced consistent lecture styles throughout
the semester, even with peer feedback. The absence of new and interactive elements in their
learning experience could have reduced their sense of inspiration and self-improvement. This
finding aligns with research on AI-mediated language instruction, which has been shown to
improve L2 motivation (Wei, 2023), and it replicates the findings of Huang and Mizumoto
(2024a), where the use of ChatGPT enhanced students’ Ideal L2 Self motivation. This, in turn,
supports the idea that AI can foster more engaging and personalized learning experiences, lead-
ing to increased motivation.
In addressing RQ2, Ought-to L2 Self, showed no significant interaction between the two fac-
tors. This indicates that ChatGPT use does not significantly impact Ought-to L2 Self. While
both groups reported slightly higher motivation in this area after participating in peer feed-
back or instructor-led GenAI-supported classes, the difference was not statistically significant
between the two groups. Therefore, the answer to this RQ is “No, students do not report higher
Ought-to L2 Self motivation after instructor-led GenAI usage in the classroom.”
The Ought-to L2 Self represents external pressure or obligations in language learning, often
seen as a sense of duty (Dörnyei, 2009). Although both groups showed slight increases in scores,
these changes were not significant. A possible explanation is that all students were enrolled to
meet mandatory credit requirements for graduation. This constant obligation to fulfill credit
requirements, regardless of tool use, aligns with findings on the Ought-to L2 Self being influ-
enced by instrumental factors (Taguchi et al., 2009). Thus, external pressures remained con-
stant and did not significantly affect language-learning motivation, which aligns with previous
findings that ChatGPT use does not impact students’ Ought-to L2 Self motivation.
For RQ3, L2 Learning Experience, the importance of interaction between the two factors is
highlighted. While students in the treatment group reported higher mean scores, the control
group showed a slight decline. This suggests that instructor-led AI use in EFL writing classes
positively impacts students’ L2 Learning Experience and enhances it. Similar to RQ1, the
answer to this RQ is “Yes, students do report a higher L2 Learning Experience motivation after
instructor-led GenAI usage in the classroom.”
The L2 Learning Experience concept involves attitudes toward learning closely associated
with the immediate learning environment (Dörnyei, 2009). In this study, the treatment group’s
mean score was slightly higher, while the control group exhibited a lower mean score. These
results are consistent with expectations, as the introduction of the tool aligns with students’
current learning context, adding a new dimension to their environment, which aligns with
Dörnyei’s (2009) construct. For the treatment group, the tool’s integration enriched their learn-
ing environment, leading to an elevated mean score. Meanwhile, the control group experienced
a steady lecture style without new elements, resulting in a lower mean score. This finding also

Digital Applied Linguistics, Volume 1 (2024)

14 Examining the effect of generative AI on students’ motivation and writing self-efficacy

supports previous research, demonstrating that ChatGPT positively influenced students’ moti-
vation in their L2 Learning Experience. Furthermore, this result can be attributed to the ability
of AI tools to personalize learning and enhance the learning environment. As noted by Leong et
al. (2024) and Yamaoka (2024), AI tools can tailor learning and function as personalized teach-
ing assistants, fostering greater motivation and a more positive learning experience.
In addressing RQ4, two significant interactions were found between the two factors. Although
both the control and treatment groups showed higher mean scores in the post-measurement,
these two significant results further validate the tool’s effectiveness. Therefore, it can be inferred
that using GenAI positively impacts students’ writing self-efficacy. Like in RQ1 and RQ3, the
answer to this question is “Yes, students report higher writing self-efficacy following instruc-
tor-led GenAI use in the classroom.”
The effectiveness of the GenAI tool in this study demonstrated how it can create a supportive
classroom environment. By providing immediate feedback, GenAI eliminates the usual time
lag students experience when waiting for feedback from teachers. Given that self-efficacy can
help reduce foreign language anxiety (Teng, 2024a), these findings reinforce the value of GenAI
as a contextual support tool that fosters self-efficacy (Teng, 2024c). Additionally, the study’s
findings align with Teng and Yang’s (2022) conclusions on the role of GenAI in enhancing both
self-efficacy and motivation. Therefore, incorporating GenAI into the classroom holds potential
for creating a more supportive learning context.
To address RQ5, the correlations of writing self-efficacy with motivational factors among stu-
dents using GenAI showed increases across all three factors. In contrast, students using peer
feedback showed either stable or decreased correlations. This suggests that using GenAI fur-
ther strengthens the relationship between writing self-efficacy and motivation. Therefore, the
answer to this question is: “Yes, the correlations of writing self-efficacy with motivational fac-
tors increase after using GenAI in the language classroom.”
Self-efficacy not only refers to students’ belief in their ability to achieve specific goals (Ban-
dura, 1997), but it also affects their performance and outcomes (Bandura, 1997; Schunk, 1991).
In this study, students demonstrated enhanced writing self-efficacy, illustrating the tool’s effec-
tiveness when implemented appropriately. Furthermore, self-efficacy and motivation are highly
correlated (Teng & Wu, 2023), so an increase in one tends to result in an increase in the other.
Referring to the first three research questions, all three motivational factors showed increased
mean scores, which aligns with the rise in writing self-efficacy. Additionally, this study demon-
strated that the correlations between writing self-efficacy and motivation strengthened after
the use of GenAI, further underscoring the tool’s positive impact. This finding reinforces the
correlation between motivation and self-efficacy reported by Teng and Wu (2023). Although
previous studies, such as those by Huang and Mizumoto (2024), did not explore this specific
aspect, the results align with expectations given the relationship between these two variables.
This evidence suggests that the careful introduction of GenAI in a classroom setting can enhance
students’ motivation and self-efficacy simultaneously.
This study, along with previous studies, provides valuable insights into the pedagogical appli-
cations of GenAI, particularly in EFL writing classrooms. The findings suggest that instruc-
tor-led integration of tools like ChatGPT can be instrumental in fostering students’ Ideal L2
Self-motivation, their perceived L2 Learning Experience, and their overall writing self-efficacy.
This implies that educators should carefully consider incorporating these AI tools into their
teaching practices, focusing on enhancing the learning environment and providing robust aca-
demic support. However, it’s crucial to note that while AI can boost intrinsic motivation, it
doesn’t significantly impact students’ Ought-to L2 Self-motivation, which stems from exter-
nal pressures and obligations. This emphasizes that AI tools are most effective when used to

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 15

foster a genuine interest in learning rather than to fulfill external requirements. Additionally,
the study advocates for a structured approach to using ChatGPT, including crafted prompts,
preliminary paper-based tasks, and documentation of changes, to prevent overreliance and
plagiarism (Huang, 2023). This underscores the need for educators to guide students towards
responsible and effective AI usage in language learning.
This study aimed to validate previous findings, and while it did confirm all of them, sev-
eral limitations remain. First, the study’s generalizability is limited because it was conducted
only in Japan with primarily first- and second-year undergraduate students. Second, the sur-
vey relied heavily on self-reported data, which can introduce potential bias. Third, students’
ChatGPT usage outside the classroom was neither monitored nor accounted for, potentially
impacting results. Fourth, data were collected over a single semester, limiting insights into
long-term effects. Fifth, the study focused solely on writing classes, so the findings may not
apply to other skills, such as speaking. Lastly, students’ actual writing proficiency was not
measured.
To address these limitations, future research should expand sample pools to include a more
diverse population across different cultural and educational backgrounds, improving general-
izability. Including students from various years and disciplines could also provide a broader
perspective on the tool’s effectiveness. Adding student interviews may further strengthen
self-report measures. Although it may be challenging, tracking students’ ChatGPT usage out-
side the classroom could yield insights into independent usage patterns, helping to differenti-
ate classroom-based effects from external influences. Conducting longitudinal studies across
multiple semesters would allow researchers to examine the effects of sustained GenAI exposure
on students’ motivation, self-efficacy, and skill development over time. Extending the study to
speaking classes and comparing effects across different language skills could offer insights into
the transferability of these tools. Finally, evaluating students’ writing samples for complexity,
accuracy, and fluency would provide more objective measures of improvement. By addressing
these limitations, future research can yield a more comprehensive understanding of GenAI’s
role in language learning and offer more robust evidence of its impact on motivation and
self-efficacy.

Conclusion
This study presents several pedagogical implications for EFL instructors integrating GenAI
tools, such as ChatGPT, into writing classrooms. First, fostering motivation and a positive
learning experience is essential. To encourage intrinsic motivation, instructors should frame AI
as a tool that empowers learning, design activities that promote active engagement and explo-
ration, and highlight ChatGPT’s potential to provide personalized feedback. Second, develop-
ing critical thinking skills alongside AI proficiency is crucial. Instructors should position AI as
a collaborative tool, not a substitute for critical thought. This can be achieved by integrating
paper-based writing and reflective activities, and by promoting responsible, ethical AI usage.
Third, creating a supportive and inclusive learning environment is key. This involves offering
structured guidance, facilitating peer collaboration, and addressing equity and access to ensure
all students can engage with AI meaningfully. Finally, promoting ongoing professional develop-
ment is essential. Instructors should be encouraged to stay informed about AI advancements
and to experiment with various AI tools and instructional approaches, which will help them
better support their students in a rapidly evolving tech landscape.
In conclusion, this study confirms the potential benefits of GenAI in EFL education, particularly
in enhancing students’ internal motivation and writing self-efficacy and strengthening their
correlations. While AI tools can transform the language-learning experience by supporting

Digital Applied Linguistics, Volume 1 (2024)

16 Examining the effect of generative AI on students’ motivation and writing self-efficacy

students’ Ideal L2 Self and enhancing their learning context, their role in affecting obligation-
driven motivation is limited. Future research should address the study’s limitations by
broadening the sample and extending research to other skill areas and learning contexts. These
findings advocate for a balanced and carefully monitored integration of AI in classrooms, aimed
at supporting authentic, sustained interest in language learning.

References
Ahmetovic, E., Becirovic, S., Dubravac, V., & Brdarevic-Celjo, A. (2023). The interplay between corrective
feedback, motivation and EFL achievement in middle and high school education. Journal of Language
and Education, 9(1), 26–41. https://doi.org/10.17323/jle.2023.12663
Aizawa, I., Rose, H., Thompson, G., & Curle, S. (2020). Beyond the threshold: Exploring English
language proficiency, linguistic challenges, and academic language skills of Japanese students in an
English medium instruction programme. Language Teaching Research, 27(4), 837–861. https://doi.
org/10.1177/1362168820965510
Al-Hoorie, A. H. (2018). The L2 motivational self system: A meta-analysis. Studies in Second Language
Learning and Teaching, 8(4), 721–754. https://doi.org/10.14746/ssllt.2018.8.4.2
Al-Hoorie, A. H., Hiver, P., & In’nami, Y. (2023). The validation crisis in the L2 motivational self system
tradition. Studies in Second Language Acquisition, 46(2), 307–329. https://doi.org/10.1017/
s0272263123000487
Allen, T. J., & Mizumoto, A. (2024). ChatGPT over my friends: Japanese EFL learners’ preferences for
editing and proofreading strategies. RELC Journal. https://doi.org/10.1177/00336882241262533
Athanassopoulos, S., Manoli, P., Gouvi, M., Lavidas, K., & Komis, V. (2023). The use of ChatGPT as a
learning tool to improve foreign language writing in a multilingual and multicultural classroom.
Advances in Mobile Learning Educational Research, 3(2), 818–824. https://doi.org/10.25082/
AMLER.2023.02.009
Bandura, A. (1997). Self-efficacy: The exercise of control. W H Freeman.
Barrett, A., & Pack, A. (2023). Not quite eye to AI: Student and teacher perspectives on the use of generative
artificial intelligence in the writing process. International Journal of Educational Technology in
Higher Education, 20, 1–24. https://doi.org/10.1186/s41239-023-00427-0
Barrot, J. (2023). Using ChatGPT for second language writing: Pitfalls and potentials. Assessing Writing,
57, 100745. https://doi.org/10.1016/j.asw.2023.100745
Cai, Q., Lin, Y., & Yu, Z. (2023). Factors influencing learner attitudes towards ChatGPT-assisted language
learning in higher education. International Journal of Human-Computer Interaction, 40(22), 7112–
7126. https://doi.org/10.1080/10447318.2023.2261725
Cen, Y., & Zheng, Y. (2023). The motivational aspect of feedback: A meta-analysis on the effect of different
feedback practices on L2 learners’ writing motivation. Assessing Writing, 59, 100802. https://doi.
org/10.1016/j.asw.2023.100802
Cotton, D., Cotton, P. A., & Shipway, J. R. (2023). Chatting and cheating: Ensuring academic integrity in
the era of ChatGPT. Innovations in Education and Teaching International, 61(2), 228–239. https://
doi.org/10.1080/14703297.2023.2190148
Csizér, K., & Kormos, J. (2009). Learning experiences, selves and motivated learning behaviour: A
comparative analysis of structural models for Hungarian secondary and university learners of English.
In Z. Dörnyei & E. Ushioda (Eds), Motivation, language identity and the L2 self (pp. 98–117).
Multilingual Matters.
Dörnyei, Z. (2005). The psychology of the language learner: Individual differences in second language
acquisition. Lawrence Erlbaum.
Dörnyei, Z. (2020). Innovations and challenges in language learning motivation. Routledge.
Fang, Y. (2023). Exploring the effects of teachers’ feed up and feed forward on Chinese senior high
students’ English writing self-efficacy and writing motivation. Journal of Education Humanities and
Social Sciences, 13, 30–36. https://doi.org/10.54097/ehss.v13i.7851
Guo, K., & Wang, D. (2023). To resist it or to embrace it? Examining ChatGPT’s potential to support
teacher feedback in EFL writing. Education and Information Technologies, 29, 8435–8463. https://
doi.org/10.1007/s10639-023-12146-0

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 17

Hao, Z., Fang, F., & Peng, J. (2024). The integration of AI technology and critical thinking in English major
education in China: opportunities, challenges, and future prospects. Digital Applied Linguistics, 1,
2256. https://doi.org/10.29140/dal.v1.2256
Henry, A., & Liu, M. (2024). Jingle-Jangle fallacies in L2 motivational self system research: A response
to Al-Hoorie et al. (2024). Applied Linguistics, 45(4), 738–746. https://doi.org/10.1093/applin/
amae041
Huang, J. (2023). Engineering ChatGPT prompts for EFL writing classes. International Journal of TESOL
Studies, 5(4), 73–79. https://doi.org/10.58304/ijts.20230405
Huang, J., & Mizumoto, A. (2024a). The effects of generative AI usage in EFL classrooms on the L2
motivational self system. Education and Information Technologies. https://doi.org/10.1007/
s10639-024-13071-6
Huang, J., & Mizumoto, A. (2024b). Examining the relationship between the L2 motivational self
system and technology acceptance model post ChatGPT introduction and utilization. Computers and
Education: Artificial Intelligence, 100302. https://doi.org/10.1016/j.caeai.2024.100302
Jeon, J., & Lee, S. (2023). Large language models in education: A focus on the complementary relationship
between human teachers and ChatGPT. Education and Information Technologies, 28, 15873–15892.
https://doi.org/10.1007/s10639-023-11834-1
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G.,
Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet,
O., Sailer, M., Schmidt, A., Seidel, T., & Kasneci, G. (2023). ChatGPT for good? On opportunities and
challenges of large language models for education. Learning and Individual Differences, 103, 102274.
https://doi.org/10.1016/j.lindif.2023.102274
Kohnke, L., Moorhouse, B. L., & Zou, D. (2023). ChatGPT for language teaching and learning. RELC
Journal, 54(2), 537–550. https://doi.org/10.1177/00336882231162868
Lai, C. Y., Cheung, K. Y., & Chan, C. S. (2023). Exploring the role of intrinsic motivation in ChatGPT
adoption to support active learning: An extension of the technology acceptance model. Computers and
Education Artificial Intelligence, 5, 100178. https://doi.org/10.1016/j.caeai.2023.100178
Leong, J., Pataranutaporn, P., Danry, V., Perteneder, F., Mao, Y., & Maes, P. (2024). Putting things into
context: Generative AI-enabled context personalization for vocabulary learning improves learning
motivation. Proceedings of the CHI Conference on Human Factors in Computing Systems (pp. 1–15).
Association for Computing Machinery (ACM).
Marzuki, Widiati, U., Rusdin, D., Darwin, & Indrawati, I. (2023). The impact of AI writing tools on the
content and organization of students’ writing: EFL teachers’ perspective. Cogent Education, 10, 1–17.
https://doi.org/10.1080/2331186X.2023.2236469
McDonald, N., Johri, A., Ali, A., & Hingle, A. (2024). Generative artificial intelligence in higher education:
Evidence from an analysis of institutional policies and guidelines. arXiv. https://doi.org/10.48550/
arxiv.2402.01659
Mizumoto, A., & Eguchi, M. (2023). Exploring the potential of using an AI language model for automated
essay scoring. Research Methods in Applied Linguistics, 2(2), 100050. https://doi.org/10.1016/j.
rmal.2023.100050
Mizumoto, A., Shintani, N., Sasaki, M., & Teng, M. F. (2024). Testing the viability of ChatGPT as a
companion in L2 writing accuracy assessment. Research Methods in Applied Linguistics, 3(2), 100116.
https://doi.org/10.1016/j.rmal.2024.100116
Mohamed, A. M. (2024). Exploring the potential of an AI-based chatbot (ChatGPT) in enhancing English
as a foreign language (EFL) teaching: Perceptions of EFL faculty members. Education and Information
Technologies, 29, 3195–3217. https://doi.org/10.1007/s10639-023-11917-z
Pajares, F., & Valiante, G. (1999). Grade level and gender differences in the writing self-beliefs of middle
school students. Contemporary Educational Psychology, 24(4), 390–405. https://doi.org/10.1006/
ceps.1998.0995
Papi, M., & Teimouri, Y. (2024). Manufactured crisis: A response to Al-Hoorie et al. (2024). Studies in
Second Language Acquisition. https://doi.org/10.1017/s0272263124000494
Ryan, S. (2009). Self and identity in L2 motivation in Japan: The ideal L2 self and Japanese learners of
English. In Z. Dörnyei & E. Ushioda (Eds.), Motivation, language identity and the L2 self (pp. 120–
143). Multilingual Matters. https://doi.org/10.21832/9781847691293-007

Digital Applied Linguistics, Volume 1 (2024)

18 Examining the effect of generative AI on students’ motivation and writing self-efficacy

Schunk, D. H. (1991). Self-efficacy and academic motivation. Educational Psychologist, 26(3–4), 207–
231. https://doi.org/10.1080/00461520.1991.9653133
Stapleton, P., & Leung, B. K. K. (2019). Assessing the accuracy and teachers’ impressions of Google
Translate: A study of primary L2 writers in Hong Kong. English for Specific Purposes, 56, 18–34.
https://doi.org/10.1016/j.esp.2019.07.001
Song, C., & Song, Y. (2023). Enhancing academic writing skills and motivation: assessing the efficacy of
ChatGPT in AI-assisted language learning for EFL students. Frontiers in Psychology, 14, 1260843.
https://doi.org/10.3389/fpsyg.2023.1260843
Taguchi, T., Magid, M., & Papi, M. (2009). The L2 motivational self system among Japanese, Chinese
and Iranian learners of English: A comparative study. In Z. Dörnyei & E. Ushioda (Eds.), Motivation,
language identity and the L2 self (pp. 66–97). Multilingual Matters.
Teng, M. F. (2023). Scientific writing, reviewing, and editing for open-access TESOL journals: The
role of ChatGPT. International Journal of TESOL Studies, 5(1), 87–91. https://doi.org/10.58304/
ijts.20230107
Teng, M. F. (2024a). Do self-efficacy belief and emotional adjustment matter for social support and anx-
iety in online English learning in the digital era? Digital Applied Linguistics, 1, 2227. https://doi.
org/10.29140/dal.v1.2227
Teng, M. F. (2024b). A systematic review of ChatGPT for English as a foreign language writing:
Opportunities, challenges, and recommendations. International Journal of TESOL Studies, 6(3),
36–57. https://doi.org/10.58304/ijts.20240304
Teng, M. F. (2024c). “ChatGPT is the companion, not enemies”: EFL learners’ perceptions and experiences
in using ChatGPT for feedback in writing. Computers and Education: Artificial Intelligence, 7, 100270.
https://doi.org/10.1016/j.caeai.2024.100270
Teng, M. F., & Wu, J. G. (2023). An investigation of learners’ perceived progress during online education:
Do self-efficacy belief, language learning motivation, and metacognitive strategies matter? The Asia-
Pacific Education Researcher, 33(2), 283–295. https://doi.org/10.1007/s40299-023-00727-z
Teng, M. F., & Yang, Z. (2022). Metacognition, motivation, self-efficacy belief, and English learning
achievement in online learning: Longitudinal mediation modeling approach. Innovation in Language
Learning and Teaching, 17(4), 778–794. https://doi.org/10.1080/17501229.2022.2144327
Tseng, W., & Warschauer, M. (2023). AI-writing tools in education: If you can’t beat them, join them.
Journal of China Computer-Assisted Language Learning, 3(2), 258–262. https://doi.org/10.1515/
jccall-2023-0008
Van Horn, K. (2024). ChatGPT in English language learning: Exploring perceptions and promoting
autonomy in a university EFL context. Teaching English as a Second Language Electronic Journal
(TESL-EJ), 28(1). https://doi.org/10.55593/ej.28109a8
Wei, L. (2023). Artificial intelligence in language instruction: impact on English learning achievement,
L2 motivation, and self-regulated learning. Frontiers in Psychology, 14. https://doi.org/10.3389/
fpsyg.2023.1261955
Xiao, Y., & Zhi, Y. (2023). An exploratory study of EFL learners’ use of ChatGPT for language
learning tasks: Experience and Perceptions. Languages, 8(3), 212. https://doi.org/10.3390/
languages8030212
Yamaoka, K. (2024). ChatGPT’s motivational effects on Japanese university EFL learners: A qualitative
analysis. International Journal of TESOL Studies, 6(3), 24–35. https://doi.org/10.58304/
ijts.20240303
Yang, S., Liu, Y., & Wu, T. (2024). ChatGPT, a new “Ghostwriter”: A teacher-and-students poetic
autoethnography from an EMI academic writing class. Digital Applied Linguistics, 1, 2244. https://
doi.org/10.29140/dal.v1.2244
Zhang, J., Zhang, L. J., & Zhu, Y. (2023). Development and validation of a genre-based second lan-
guage (L2) writing self-efficacy scale. Frontiers in Psychology, 14. https://doi.org/10.3389/
fpsyg.2023.1181196
Zheng, Y., Wang, Y., Liu, K. S., & Jiang, M. Y. (2024). Examining the moderating effect of motivation
on technology acceptance of generative AI for English as a foreign language learning. Education and
Information Technologies. https://doi.org/10.1007/s10639-024-12763-3

Digital Applied Linguistics, Volume 1 (2024)

Examining the effect of generative AI on students’ motivation and writing self-efficacy 19

Author Bio
Jerry Huang is an English language instructor at Kansai University and originally from Los
Angeles. He holds a master’s degree from Hyogo University of Teacher Education and is a Ph.D.
candidate in foreign language education and research at Kansai University. Jerry’s teaching
expertise lies in English language courses, and his research interests include motivation, AI,
ChatGPT, and Languages Other Than English (LOTE).

Atsushi Mizumoto holds a Ph.D. in Foreign Language Education and is a professor in the Fac-
ulty of Foreign Language Studies and the Graduate School of Foreign Language Education and
Research at Kansai University, Japan. His current research interests include corpus use for
pedagogical purposes, learning strategies, language testing, and research methodology.

Digital Applied Linguistics, Volume 1 (2024)

20 Examining the effect of generative AI on students’ motivation and writing self-efficacy

Appendix

Ideal L2 Self

• I can imagine myself living abroad and having a discussion in English.

• Whenever I think of my future career, I imagine myself using English.
• I can imagine a situation where I am speaking English with foreigners.
• I imagine myself as someone who is able to speak English.
• The things I want to do in the future require me to use English.

Ought-to L2 Self

• I study English because close friends of mine think it is important.

• Learning English is necessary because people surrounding me expect me to do so.
• I have to study English, because, if I do not study it, I think my parents will be disap-
pointed with me.
• My parents believe that I must study English to be an educated person.

L2 Learning Experience

• I like the atmosphere of my English classes.

• I always look forward to English classes.
• I find learning English really interesting.
• I really enjoy learning English.

Writing Self-Efficacy

• I can correctly spell all words in my paragraph.

• I can correctly punctuate my paragraph.
• I can correctly use all parts of speech in my paragraph.
• I can write simple sentences with good grammar.
• I can correctly use singulars and plurals, verb tenses, prefixes, and suffixes.
• I can write a strong paragraph that has a good topic sentence and main idea.
• I can structure paragraphs to support ideas in the topic sentences.
• I can end paragraph with proper conclusions.
• I can get ideas across in a clear manner by staying focused without getting off topic.

Digital Applied Linguistics, Volume 1 (2024)

View publication stats