Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 157

Exploring the Utility of ChatGPT for Self-Directed

Online Language Learning
Zixi Li
Indiana University Bloomington, USA
[Link]

Dr. Chaoran Wang

Colby College, USA
[Link]

Dr. Curtis J. Bonk

Indiana University Bloomington, USA
[Link]

Abstract

As generative AI tools are increasingly popular in today’s teaching and learning process,
challenges and opportunities occur at the same time. Self-directed learning has been regarded as
a powerful learning ability that supports learners in informal learning contexts and its importance
rises in salience when incorporating AI into learning. This study employed a mixed-method
design to understand how people self-direct their online language learning through the utilization
of ChatGPT. Analyzing survey data from 276 survey respondents and 11 one-to-one interviews
with language learners in the United States, we found that learners are motivated to use
generative AI for its high flexibility and personalization which enables learners to access
learning materials that align with their knowledge levels, personal interests, and learning goals.
We also found self-monitoring skills that are inherent to learners help them to use ChatGPT
more effectively and achieve their learning goals efficiently. This study contributes to the
emerging field of AI in education and explores possibilities to use ChatGPT to foster self-
directed language learning and provide educators, instructional designers, and researchers with
insights to design learning integrated with AI to best fulfill learners’ diverse needs and expand
learning opportunities to more people.

Keywords: Self-directed learning, language learning, artificial intelligence in education,

generative AI, ChatGPT

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 158

Li, Z., Wang, C., & Bonk, C.J. (2024). Exploring the utility of ChatGPT for self-directed online
language learning. Online Learning, Volume28(3), (157-180). DOI:10.24059/olj.v28i3.4497

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 159

The integration of artificial intelligence (AI) tools in teaching and learning has
demonstrated its effectiveness in alleviating learner anxiety while elevating their confidence and
fostering increased motivation for communication (Bao, 2019; Tai & Chen, 2020). In recent
years, generative AI (GenAI) technology has significantly grown, particularly in educational
settings (Baidoo-Anu & Ansah, 2023; Chan & Hu, 2023). GenAI features utilizing machine
learning technologies to analyze patterns of existing data to generate new human-like content in
formats including but not limited to text, visuals, and audio (Brynjolfsson et al., 2023).

ChatGPT is currently one of the most known and used GenAI platforms with its
remarkable capacity to complete complex tasks and wide implementation in education (Adeshola
& Adepoju, 2023; Baidoo-Anu & Ansah, 2023; Rahman & Watanobe, 2023; Yu & Guo, 2023).
Recent research indicates that GenAI tools and platforms, such as ChatGPT, confer advantages
to language learning through the generation of coherent and fluent text, the creation of authentic
language materials, and the facilitation of personalized learning (Baskara, 2023a). However,
despite these advancements, there is limited empirical research concerning the role of AI in self-
directed language learning (SDLL), particularly in informal learning settings (Jeon, 2022). To
close the knowledge gap, this research attempts to understand how adult learners self-direct their
online language learning through the utilization of GenAI platforms such as ChatGPT.

Literature Review
Generative AI-Supported Self-Directed Learning
Self-directed learning (SDL) is a central element of adult learning and has been regarded
as a powerful learning ability that strongly supports learners in informal learning contexts (Li &
Bonk, 2023; Brookfield, 2009; Morrison & McCutheon, 2019). SDL emphasizes learning
autonomy, personal responsibility, and self-growth (Gibbons, 2002; Wilcox, 1996). It is often
investigated through the perspectives of being a personal attribute, a process, or a context of the
learning environment (Song & Hill, 2007). One of the most influential SDL frameworks was
designed by Garrison (1997) over two decades ago. It recognizes SDL through the lens of (1)
motivation, (2) self-management, and (3) self-monitoring. Garrison (1997) claimed that
motivation is the internal desire or need that leads individuals to engage in SDL, which includes
both entering motivation and task motivation. The focus of self-management is self-control
towards what and how to learn, including the use of resources, control of the learning context
toward the goal, and strategies to manage the learning schedule. Self-monitoring is the
responsibility for the process of learning, including being reflective and thinking critically.

Research has proven that incorporating GenAI into SDL has a positive impact on
learning (Preiksaitis & Rose, 2023). Lin (2023) suggested that ChatGPT particularly has great
potential to address the challenges faced by adult learners, such as the difficulty of locating
resources and the lack of a supportive learning environment. In addition, ChatGPT can support
learners in establishing learning objectives, identifying accessible resources, creating
personalized learning plans, tracking learning progress, reflecting on learning experiences, and
ultimately achieving successful forms of SDL, though instructor support may still be needed in

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 160

this process. Furthermore, Baidoo-Anu and Ansah (2023) proposed that ChatGPT fosters
personalized and interactive learning, thereby promoting SDL. They emphasized that ChatGPT
encourages students to independently explore learning topics by inputting keywords or questions
to access relevant information and resources (Firat, 2023; Rasul et al., 2023). While scholars
have argued that ChatGPT provides personalized learning support, Chan and Hu (2023) noted
that concerns about accuracy, privacy, and other ethical and social issues were raised by their
survey participants.

Generative AI for Language Learning

GenAI platforms have become prominent for language learning and teaching (Anggoro &
Pratiwi, 2023). Research suggests that ChatGPT could support language learning by providing
scaffolding and feedback, developing necessary language skills, and recommending activities for
language practice (Bin-Hady et al., 2023). Liu and Ma (2023) conducted a survey involving 405
Chinese undergraduate and graduate English learners to investigate their perceptions and
acceptance of using ChatGPT in supporting their English learning. Their findings show that the
ease of using ChatGPT influenced participants’ perceived usefulness of the tool, which
significantly impacted the students’ utilization of ChatGPT in informal language learning
settings. Liu and Ma (2023) suggested that ChatGPT provides easy access and increased
opportunities to engage in SDLL among language learners.

Liu (2023) also examined English learners’ attitudes towards ChatGPT and the
effectiveness of using ChatGPT to facilitate language learning. This study sampled 109 Chinese
students at the undergraduate level, including nearly 37% of participants currently enrolled in
colleges in Mainland China, around 3% from Hong Kong, Macao, or Taiwan, and the remaining
60% of students studying overseas. Adapted from a questionnaire by Hasan (2019), five-point
Likert scale questions were designed to measure perceptions of ChatGPT, attitudes towards
actual practice with ChatGPT, perceptions of the advantages of ChatGPT, and the perceptions of
disadvantages of ChatGPT. Hasan’s (2019) findings suggested that participants have a slightly
stronger tendency to use ChatGPT to improve reading and writing skills over communication
skills, and they demonstrated generally positive attitudes toward ChatGPT’s potential to boost
their learning abilities. Interestingly, this study indicated that there is a relationship between
students’ geographic location and their acceptance of ChatGPT. Those who study in Mainland
China displayed lower levels of motivation and a less positive attitude towards using ChatGPT
for English learning than students who were studying overseas. Thus, when considering how
learners use ChatGPT for SDLL, it seems likely that the learning context and social elements
such as institutional policy and sociocultural norms significantly impact their learning
experiences (Li et al., 2024). Liu and Ma (2023) and Liu (2023) shared some common research
designs such as employing quantitative surveys among Chinese students in higher education
settings. However, such quantitative approaches are not sufficient to measure learner attitudes
and perceptions.

In addition, current research on GenAI for language learning has predominately focused
on learning English as a second language in classroom settings as well as investigating

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 161

perceptions instead of practice. For instance, Lee et al. (2024) investigated the perceptions of 80
English language learners (ELLs) from a Korean university. They reported that 30% of the
participants felt uncertain about the efficacy of these tools in facilitating English learning. Lee et
al. (2024) suggested a need to investigate the possible integration of AI tools into English
language teaching for adult learners worldwide. However, apart from these perception-based and
classroom-focused language learning research studies, individual learners’ self-directed language
learning practices remain largely unexplored. Given that language learning technologies (e.g.,
Duolingo, Rosetta Stone) have been long and widely used in SDL (Li & Bonk, 2023), research
should strive to understand how new generative AI technologies can impact self-directed
language learning.

In this study, we investigated the SDL practices facilitated by GenAI tools like ChatGPT
following Garrison’s (1997) SDL framework. Aligning with each dimension of Garrison’s
framework, namely, motivation, self-management, and self-monitoring, we examined the
following research questions accordingly:

1. What motivates learners to utilize self-directed learning with ChatGPT?

2. How do language learners select and apply self-management strategies when utilizing
ChatGPT in their language learning routines?
3. How do language learners employ self-monitoring strategies when using ChatGPT for
language learning?

Methods

This study aims to fill a prominent gap in the literature by investigating how language
learners incorporate ChatGPT into their SDLL through an explanatory mixed-method approach
(Creswell & Plano Clark, 2007). Drawing on Garrison's (1997) self-directed learning (SDL)
model, the present study developed survey and interview instruments to measure SDL across
three dimensions: (1) motivation, (2) self-management, and (3) self-monitoring.

Data Collection
Survey. The survey consisted of three demographic questions, five inquiries on general
ChatGPT usage, and a set of three Likert scale questions with 23 items tailored to assess learners'
SDL in utilizing ChatGPT for language learning. The survey items were adapted from
instruments by Fisher and King (2010), Williamson (2007), and the Zhu and Bonk (2019), which
are based on Garrison’s (1997) SDL framework. As we developed the survey instrument, we had
one SDL expert to check on items and a language learner to pilot this survey. We made changes
accordingly after we gathered feedback. Example survey questions can be found in Table 1.

Table 1
Examples of Survey Questions and Aligning SDL Principles

Items Strongly Agree Neither agree Disagree Strongly

agree nor disagree disagree

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 162

Motivation (Note: this title is not included in the actual survey)

I have motivation to learn a language

with ChatGPT.
I enjoy learning new information
related to language through
ChatGPT.

Self-management

I seek assistance when facing

problems learning a language online.
I manage my time well while using
ChatGPT to learn a language online.
Self-monitoring

I can direct my own language

learning progress while learning with
ChatGPT.
I have high learning standards when
I learn a language with ChatGPT.

A survey distribution platform, Prolific, was utilized for participant recruitment, targeting
individuals who met the following specific criteria: at least 18 years old, residing in the United
States, currently enrolled as a student, and having engaged with ChatGPT for second language
learning purposes.

A total of 276 completed survey responses were collected. Among all the survey
participants, 33.45% identified themselves as female, 64.36% as male, and 2.18% as other (e.g.,
non-binary). In terms of educational levels, 24.73% of participants have their highest degree
from high schools, 49.09% have bachelor’s degrees, 13.82% have master’s degrees, 1.45% have
doctoral degrees, and 10.91% have other types of highest degrees (e.g., associate degrees,
technical degrees). As Figure 1 indicated, 136 (49.28%) of the survey respondents reported that
they used ChatGPT to facilitate their language learning weekly, followed by 66 (23.91%) on a
monthly basis, 45 (16.30%) rarely, and 29 (10.51%) daily.

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 163

Figure 1
Frequency Of Using ChatGPT To Facilitate Language Learning.

Interviews. At the end of the survey, participants were asked if they were willing to
participate in a one-time 40-minute interview through a Zoom meeting room of the researchers.
If so, the respondents would provide their contact email address via the survey form. After
collecting their email addresses (N=125), we used random sampling to select interview
participants. Those selected would receive an interview invitation through email with an
information sheet about the study. Those willing to engage in the interview would reply to the
email invitation and confirm their participation. We contacted all 125 participants who left their
contact information, and, eventually, 11 individuals agreed and participated in interviews. Their
demographic information is included in Table 2.

Table 2
Demographic Information Of Participants

Gender Frequency of using

Participant Current education ChatGPT for language
pseudonym Age level Target language learning

Amelia 31 Female Graduate degree Spanish Weekly

Bennett 31 Male Graduate degree Spanish Daily

Chandler 41 Male Graduate degree Spanish Weekly

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 164

Danielle 29 Female Undergraduate Hebrew Weekly

Elliott 35 Male Graduate degree Spanish Weekly

Fallon 40 Female Undergraduate Spanish Weekly

Giovanni 45 Male Undergraduate Spanish Daily

Howard 24 Male Undergraduate French Weekly

Male Brazilian
Ivan 41 Graduate degree Portuguese Weekly

Male Spanish, Italian,

Kai 23 Undergraduate Turkish, French Weekly

Lydia 48 Female Graduate degree Tamil Daily

Artifacts. We either asked them to share their screens during the interview to walk us
through their typical use of ChatGPT for language learning or send screenshots of previous
language learning related interactions with ChatGPT.

Data Analysis

The analysis employed descriptive statistics for survey data (Mishra et al., 2019).
Researchers used various graphical representations to summarize and indicate the distribution,
central tendency, and variability among responses.

To systematically analyze data and identify recurring patterns, we used thematic analysis
for interview data analysis (Braun & Clarke, 2022). We started with familiarizing ourselves with
interview data and preparing them for analysis by transcribing all the audio recordings
automatically and then re-listening to make corrections manually. During the first stage, we
created initial codes to represent meanings and patterns that were observed across the data. In
NVivo 14, we highlighted the excerpts and assigned appropriate codes to them, followed by
reading through all excerpts within each code to adjust and revise codes as needed. Then, we
grouped codes into themes that have nuanced and meaningful information about the research
questions.

During the coding process, to reduce individual biases and increase the validity and
reliability of the coding process, two researchers conducted a pilot test where each of them coded
two interviews before coding the entire dataset. Then, when conducting individual coding, these
researchers used a blind coding scheme where they were not informed of each other’s coding

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 165

decisions until the end, which minimizes the potential biases. Next, they compared coding results
and openly discussed any discrepancies to refine coding guidelines to ensure agreement between
coders. They also had regular meetings to discuss if any challenges or discrepancies occurred.
Table 3 delineates the coding themes, codes, and examples of excerpts.

Table 3
Coding Themes, Codes, And Examples of Excerpts

Themes Codes Examples of excerpts

Motivation Initially motivated by “I talked to a friend who uses AI in place of

people who used actually speaking to a person one-on-one
ChatGPT in the when learning a language. And she was
surrounding talking about how her experiences are really
environment positive, and she's learned a lot from it. And
so I decided to kind of explore that and like,
see how it goes.”

Motivated by “But if I have a question where I'm stuck, I

ChatGPT’s convenience can ask ChatGPT. I don't have to rely on a
for supporting study partner who may or may not know the
independent learning answer.”

ChatGPT makes “It's more contextual, like you can tell him,
learning efficient by ‘Hey I have a friend. We're not too close. I
offering contextualized want to say this, how do I say it?’ Like, it's
and point-to-point the context behind it.”
answers

Motivated by “I can always ask how this grammar was

ChatGPT’s rich structured, or why is this and not that, or why
linguistic affordances is this word instead of that word?”

Self-management Set habitual learning “I usually use it in the morning before I start
routines working. And I'll usually use it for like 10 or
15 min or so [to] have a short conversation
before I work.”

Prompt questions to “I have found that the more detailed your

align with learning goals question is, the better your answer and your

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 166

outcome is.”

Validate information “I actually have a dictionary to consult if the

with other resources word is correct… I tend to even consult
throughout the learning outside experts, I tend to use those two, [or]
process three things the most.”

Self-monitoring Know what aspects of “So what I do with it mostly is try to validate
language learning need some of what I have… the way I'll use
to be improved ChatGPT is to look for better ways to write
something.”

Compare the learning “If I walk away from it feeling like I learned
outcome with past something… that's a bit of a victory that I'm
learning experiences doing well, and I'm learning something.”

Integrate multiple “I'll just start talking [to family members who
resources or tools for speak Spanish] and they'll respond. So that
assessment shows me where I'm at at that point.”

Researchers applied document analysis for artifacts—participants’ chat history—to analyze the
language use, interaction dynamics, and patterns with ChatGPT (Bowen, 2009). This analysis of
participants’ interactions with ChatGPT was used to triangulate with our survey and interview
data.

Findings

Motivation
As shown in Table 4, the highly-ranked motivation from survey respondents is the
enjoyment of learning new information through ChatGPT. The highest-ranked statement, “I
enjoy learning new information related to language through ChatGPT,” has a mean of 4.21 and a
standard deviation of 0.78. The least voted motivation from the survey is “I like to share my
ChatGPT language learning experiences with others” (M= 3.37, SD = 1.17).

Table 4
Descriptive Statistics on Motivation Items

Statement item Mean SD Variance

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 167

I enjoy learning new information related to language through 4.21 0.78 0.61
ChatGPT.

I have motivation to learn a language with ChatGPT. 4.01 0.93 0.86

I want to learn new ideas related to language learning brought 3.94 0.99 0.98
up by ChatGPT.

I have a need to learn a language with ChatGPT. 3.43 1.22 1.49

I need to know the deeper reasons for the language feedback by 3.43 1.13 1.28
ChatGPT.

I like to share my ChatGPT language learning experiences with 3.37 1.17 1.37
others.

Interestingly, even though this motivation is not significant from survey results,
interviews found that many participants’ entering motivation to use ChatGPT for language
learning derived from the positive learning experiences indicated by people in their surrounding
environments (e.g., friends, professors, social media users, etc.). For example, Amelia’s friend
shared her language learning experience with AI platforms which inspired Amelia to try
ChatGPT. She mentioned that,

I talked to a friend who uses AI in place of actually speaking to a person one-on-one

when learning a language. And she was talking about how her experiences are really
positive, and she's learned a lot from it. And so I decided to kind of explore that and like,
see how it goes… and it's gone really well, I feel like I've learned a good amount and that
coupled with my language learning apps that I use and have been using for a couple of
years. I feel like it helps a good amount.

Danielle also started to use ChatGPT for language learning after she perceived its benefits in a
class where her professor suggested learning with ChatGPT and even provided guidance on how
to use it effectively. Notably, social discussions on ChatGPT’s potential in facilitating learning
also influenced people’s motivation to use it for language learning. For instance, Elliott indicated
that the reputation of AI tools motivated him to try ChatGPT for language learning.

The most mentioned motivation by interviewees was that ChatGPT supports convenient
and independent learning in various ways. Ivan, a graduate student who is learning Portuguese
with ChatGPT, suggested that he preferred self-study at his own pace and ChatGPT facilitated
this style of learning so that his learning was not dependent on other people. As he explained,

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 168

I typically don't like to do group study sessions or like a lot of one-on-one learning
sessions. I like to study by myself and just work and study at my own pace. Like
ChatGPT, AI tools are very well suited for me because I can do that. But if I have a
question where I'm stuck, I can ask ChatGPT. I don't have to rely on a study partner who
may or may not know the answer. For me, in that regard, I feel like the AI tools work
very well for my approach to studying.

Similarly, Danielle said, “one of the positives to me is that I know it's not a person, and so I'm
not bothering someone.” She further conveyed feeling embarrassed when not presenting herself
effectively to other humans. However, interacting with a chatbot that does not have personal
opinions provided her with a safe learning space where she could error without fear. For some
other people, ChatGPT could be the one-to-one tutor at no cost to provide personalized learning
experiences. For instance, Amelia commented that because she did not have a friend who could
coach her language learning, “ChatGPT is kind of a stand-in for a person or a tutor that knows
the language natively.” Therefore, generative AI like ChatGPT enabled learners to be more
independent in the learning process with easy access.

A very significant motivation for people to use ChatGPT for language learning is that it is
capable of providing contextualized and point-to-point answers. Elliott commented that he had
Spanish-speaking friends so sometimes he needed to figure out what his friend was saying. One
strategy he used was to copy and paste random texts from chats to obtain translations. Elliott
explained,

I copy it from here, paste it there, see what it says, what does this mean? I can do that on
Google as well. But it doesn't give me the proper context. Or I kind of trust [ChatGPT]
more, because it [is] to the point.

He further suggested that ChatGPT enabled him to ask follow-up questions, thus making his
learning more efficient. He claimed, “I can always ask how this grammar was structured, or why
is this and not that, or why is this word instead of that word? I can ask [anything], it can further
elaborate the questions I have. With Google, it's totally different.”

As shown in Figure 2, Chandler, a Spanish language learner, demonstrated his interaction

with ChatGPT on the screen. The screenshot provided a glimpse of how he phrased his questions
and the kind of tasks he frequently asked ChatGPT for. As Chandler indicated, he asked
ChatGPT to translate words or structure sentences, which was specific to his need at that
moment.

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 169

Figure 2
A Screenshot of the Participant’s Demonstration of His Interaction with ChatGPT

Self-Management
Our survey data indicated that time management has been considered in participants’
learning process. Regarding the statement, “I manage my time well while using ChatGPT,”
nearly half of the participants (48.91%) reported that they somewhat agreed. This was followed
by 25.72% who strongly agreed, 15.94% who neither agreed nor disagreed, 7.61% who
somewhat disagreed, and a mere 2.54% who strongly disagreed. The mean value is 3.87, with a
standard deviation of 0.98.

Figure 3
Learners’ Responses to the Statement on Time Management Percentages of

Interview data reinforced that the flexibility of accessing ChatGPT anytime and anywhere
motivates individuals to establish habitual learning routines, thereby supporting effective time
management. For example, Bennett who used ChatGPT on a daily basis to study Spanish stated

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 170

that he enjoyed practicing conversations at the beginning of the day. He detailed his approach as
follows:

I usually use it in the morning before I start working. and I'll usually use it for like 10 or
15 min or so [to] have a short conversation before I work. If I have more time or I feel
like I want to go back in there and have another conversation. I'll hop back in. But it's
usually at a minimum like 10 or 15 min a day at the beginning of the day.

Giovanni who has also used ChatGPT for Spanish learning also shared his learning
schedule, stating that, “I try to study at least 30 min up to 60. And then, when I'm having a
conversation, it's usually like, maybe recently, last week, I tried maybe 4 times out of that week.
So, either by voice or text.” He further emphasized the importance of time management, while
challenges exist, particularly in staying on track with learning tasks with a busy life schedule. He
noted,

One of the things I'm lacking is the time management aspect. Because I think if I had a
specific time or schedule that I have set up to do these things. So I don't start slipping and
just putting it off. Because I do work a lot, so that gets in the way. So I think that would
be the major barrier. But that's something that ChatGPT can help you with really.

Furthermore, many interviewees indicated that an effective strategy they used to learn
language with ChatGPT was prompting good questions. Chandler, who is learning Spanish on a
weekly basis, discovered that “I have found that the more detailed your question is, the better
your answer and your outcome is.” He emphasized the importance of having the ability to
promote good questions and validate information provided by ChatGPT. As Chandler shared, “I
will sometimes try phrasing my question in a number of different ways to see what the responses
are, and it's such a trick that I use with ChatGPT in general.” Danielle has been learning Hebrew
with ChatGPT; similarly, she recommended developing pre-prompt skills throughout the
learning process with ChatGPT. She suggested including some contextual information to inquire
for more accurate answers.

As our participants indicated that they recognized inaccurate or even false information
related to language learning provided by ChatGPT, they often utilized various resources to assist
their learning and validate the information. For instance, Fallon shared her experience as follows:

I've asked the same question [to ChatGPT], like maybe two or three times and it came
back with a different answer. So sometimes I'll hit I'll put it into like the Google
Translator. I don't have any other translator on my phone. So I just have been using that
and seeing what that comes back. Most of the time. It's like, oh, it's alright. And if it
doesn't, it'll tell me what the accurate one is. And then I'll go back in and I'll correct it.

In fact, most of our interviewees reported that they had used various resources or tools, both
digital (e.g., phone applications for language learning) and non-digital (e.g., books), to support
their information validation through language learning with ChatGPT. Howard suggested, “I
actually have a dictionary to consult if the word is correct. I also use Duolingo as well. And as I
Online Learning Journal – Volume 28 Issue 3 – September 2024
 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 171

was saying earlier, I tend to even consult outside experts, I tend to use those two, three things the
most.”

Self-Monitoring
Using various resources is not only a complementary learning strategy for self-
management, but it also supports self-evaluation with diverse parameters. One interviewee,
Giovanni, used fluency in communication with family members as an indicator of his learning
improvement. He elaborated,

I've been sending those messages like I said, sometimes over the phone when I'm talking
to a family member. All of them start talking to them out of nowhere, and just try to use
some words and try to communicate, and I'll see where I failed there, and just stop. I
won't warn them ahead of time. I'll just start talking [to family members who speak
Spanish] and they'll respond. So that shows me where I'm at at that point.

Apparently, Giovanni would justify his language learning with ChatGPT approach if he has
improved his Spanish language skills based on how far the conversation in Spanish with his
family lasted and also cognitively identified where further improvement is possible. While it is
an example of how people use native speakers to examine learning progress and facilitate self-
monitoring when learning a language with ChatGPT, many other participants self-evaluate their
learning by using other strategies and tools. Among the other language learning tools people use
for self-monitoring, Duolingo, has been the most mentioned.

However, compared with other tools, participants indicated that ChatGPT has a unique
advantage for language learning due to its high-level flexibility and personalization. For
example, Elliott compared his experience with other language learning tools with ChatGPT as
follows:

Duolingo is too repetitive. It does a good job if you're trying to learn something, but for
quick things or the purposes that I have, I don't want to learn it like conversational kind
of and I'm not at that point yet. So [my learning goal] is more like, okay, what do you call
this in Spanish, or what do you call that in Spanish?

He explained that because his goal is to simply understand what a Spanish-speaking friend is
saying, he just wanted to get quick answers by copying text and translating. He noted, “copy it
from here, paste it there, and see what it says. What does this mean? I can do that on Google as
well. But it doesn't give me the proper context.” This indicates that participants think of learning
by outlining what aspects of language learning need to be improved and then match the most
effective resource or tool with it.

In addition, another interviewee pointed out that tools like Duolingo are too basic for his
level, whereas ChatGPT has been a good choice because it can meet him at the level he is and
provide learning materials that match his needs. However, interestingly, some interviewees
Online Learning Journal – Volume 28 Issue 3 – September 2024
 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 172

indicated that using other tools can complement language learning with ChatGPT because it is
harder to self-monitor their language learning processes with ChatGPT. For instance, Ivan used a
combination of Duolingo and ChatGPT for learning Portuguese. He acknowledged,

I think a ChatGPT is more of a supplement, I guess, rather than a primary language

learning tool. It's good you have you use it, double check it, use Google Translate, and
also a native speaker… I can just monitor my own progress and I just know then just this
is kind of where Duolingo comes in. But when I'm taking tests or quizzes within
Duolingo, just like monitoring my performance, kind of in that as well.

Similarly, another participant indicated that ChatGPT could not facilitate a self-
monitoring process because it did not provide metrics, so she implemented other external
tools to track her learning. As she shared,

No, I don’t really use it that way [to measure my learning process]. I usually check my
progress with the other apps. I kind of have built in quizzes or with workbooks. Like if I
can complete those exercises more easily, that's sort of my metric, and I don't really know
which thing necessarily made it better because I use them all in tandem, but all of it
helps, and it helps more to have it than not have it.

Though it seems that ChatGPT is not effective in supporting self-monitoring, learners

reported very positive results in terms of how ChatGPT improved their learning. As indicated in
Figure 4, related to the statement that “ChatGPT helped me improve my knowledge of the target
language,” 80 of 276 (28.99%) survey respondents indicated that they “strongly agree,” while
150 participants (54.35%) reported “somewhat agree.” More specifically, strong agreement was
indicated from the perspective of having a higher awareness of vocabulary usage. Therefore,
ChatGPT proves particularly effective in enhancing learners' awareness of vocabulary usage and
how vocabulary and grammar are combined to form coherent and consistent texts.

Figure 4
Percentages of Learners’ Responses to Statements on the Improvement of Target Language Learning

Furthermore, learners actually demonstrated a high level of self-control through the

survey. The statement, “I carry out my own study plan while using ChatGPT for learning a
Online Learning Journal – Volume 28 Issue 3 – September 2024
 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 173

language” has received 30% of “strongly agree” and 49% of “somewhat agree.” The statement,
“I direct my own language learning progress while learning with ChatGPT,” also received a high
agreement, with 35% indicating “strongly agree” and “54%” signaling “somewhat agree.” In
effect, the survey data, along with interview data, suggested that self-monitoring is an inherent
ability of the learners, instead of relying on ChatGPT to monitor their learning. In other words,
the survey results indicate that learners exhibit a high level of responsibility towards their own
learning.

Figure 5
Percentages of Learners’ Responses to the Statement on the Study Plan

Figure 6
Percentages of Learners’ Responses to the Statement on Self-Directing Language Learning Progress

Discussion

Existing empirical studies on ChatGPT for language learning have certain limitations
regarding the methodological approach, research scope, homogeneity of samples, language-
specific focus, and learning context. For instance, a number of existing studies have been
designed based on quantitative survey methods (e.g., Liu, 2023; Liu & Ma, 2023) to investigate
Chinese learners’ perceptions and attitudes towards using ChatGPT for learning English as a
second language. However, such an approach neglects the diversity of language learners as well
as their needs and practices in other geographical, cultural, and linguistic contexts. It is
particularly problematic when comparing language learning in formal higher education settings
with adult learning in more informal learning contexts which tend to have vastly different learner
motivation variables and learning strategies (Li & Bonk, 2023; Botero et al., 2019).
Accordingly, this study aims to expand our understanding of SDLL of adult learners that
occurs outside of classroom learning where learners take initiative and responsibility for their
Online Learning Journal – Volume 28 Issue 3 – September 2024
 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 174

own learning. Importantly, with the emergence of GenAI in education, exploring different
possibilities and strategies for incorporating ChatGPT to foster SDLL will benefit millions of
adult learners who are learning languages for diverse life needs as well as leisurely pursuits and
hobbies (Baskara, 2023b; Lin, 2023; Niyozov et al., 2023). As such, the present study revealed
the role of motivation, self-management, and self-monitoring of language learners when using
ChatGPT to facilitate their SDLL.
Motivation plays a crucial role in engaging learners with ChatGPT for language learning.
The study found that learners are motivated by the flexibility and personalization that ChatGPT
offers, allowing them to access learning materials that align with their individual needs and goals
when they are deemed necessary. The findings suggest that ChatGPT can cater to the diverse
needs of learners, offering a tailored learning experience that other digital and non-digital
learning resources cannot provide (Li et al., 2024; Mogavi et al., 2024). Furthermore, the
influence of socially shared positive experiences among peers and other social interactions or
discussion, demonstrates the importance of community and social validation in adopting new
technologies, especially controversial tools like generative AI (Hwang & Chen, 2023), for
learning.
In terms of self-management, learners who effectively manage their learning schedules,
set learning routines, and strategically pose questions to ChatGPT can enhance their learning
efficiency. The finding indicates the need for learners to possess and develop strong self-
management skills to maximize the benefits of ChatGPT in language learning. The use of
ChatGPT enables learners to practice language skills flexibly, but the effectiveness of this
practice is significantly influenced by the learner's ability to self-manage their learning process
(Lin, 2023). Additionally, self-monitoring strategies are essential for learners to evaluate their
progress and adapt their language learning strategies accordingly. The self-directed language
learners in this study used ChatGPT alongside various other resources and tools for validation
and assessment, indicating that while ChatGPT is a valuable resource for language learning, it is
part of a broader ecosystem of learning tools (Li et al., 2024). This highlights the importance of
integrating ChatGPT with other learning resources and tools to provide a comprehensive learning
experience that addresses different aspects of language learning (Al-khresheh, 2024), such as
vocabulary acquisition, grammatical understanding, translation, and conversation.
It is apparent from this study that the experience of using other language learning
resources to complement learning with ChatGPT plays an important role across the three
dimensions of SDL. Most interview participants indicated that they have previously used other
tools for language learning which enabled them to see the unique advantages of ChatGPT. For
instance, many participants compared ChatGPT with Duolingo and Google Translate, suggesting
that learners who use these tools often struggled to find features that provided hands-on practice,
timely, individualized support and sufficient learning context (Li and Bonk, 2023; Jumabekovna,
2024). By contrast, ChatGPT fosters more self-directed and independent learning as well as
contextualized point-to-point answers. Those unique benefits of ChatGPT helped form a strong
motivation for people to learn languages with it. This finding confirmed that ChatGPT can
support learners’ independent exploration and information-gathering in a personalized manner
Online Learning Journal – Volume 28 Issue 3 – September 2024
 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 175

through goal-driven inquiry and self-selected questioning (Baidoo-Anu & Ansah, 2023; Firat,
2023; Rasul et al., 2023).
While prior research suggests that ChatGPT fosters SDL by identifying accessible
resources and providing personalized learning plans (Lin, 2023), our findings present novel
insight which reveals that utilizing ChatGPT solely is not sufficient in terms of self-management
and self-monitoring. In fact, we found that language learners often self-manage multiple learning
resources to complement the use of ChatGPT, as well as combine various learning resources to
evaluate their learning with ChatGPT and validate information provided by ChatGPT. As this
study reveals, ChatGPT is one of the tools through which learners can practice and adjust their
SDL strategies to facilitate effective language learning. In other words, ChatGPT can be a useful
technology tool for SDL, but it is often not the sole factor in effective SDL, particularly in terms
of self-management and self-monitoring. Learners naturally select and apply a wide range of
strategies and resources to enhance their SDLL experience where ChatGPT is one component of
the entire SDL approach. Such findings imply that language learners often require resource
integration, and the role of ChatGPT is as a facilitator instead of a substitute for SDL. These
results suggest it is beneficial to integrate generative AI with other language learning platforms
as a means to provide personalized and adaptive learning experiences.
Limitations and Future Research
Though we emailed all 125 participants multiple times who expressed interest in being
interviewed and provided their contact information, only 11 actually participated in the study
interviews. As a result, the sample size is small. We wish to expand this sample size in future
studies to reach data saturation. Additionally, given that our participants were mostly Spanish
learners, future investigations should compare and contrast the needs of learners of different
target learning languages to understand the important nuances that might be embedded in
different cases. In addition, as we collected the interview data, we discovered that the volunteer
participants were predominantly non-traditional learners who continued or even returned to their
education at a later stage of their lives.

Future research might consider the types of backgrounds and life experiences (Knowles,
1968) that contribute to adult learning and engagement with AI-supported tools and resources.
Researchers might examine the role of previous language learning experience, technology
competency, cultural background, and educational level in the process of using ChatGPT for
language learning. By investigating the unique and diverse needs of adult learners, instructional
designers and generative AI developers may build a more tailored, innovative, and effective
learning experience with generative AI. Additionally, further research is needed to explore how
ChatGPT can be effectively combined or supplemented with other educational technologies and
pedagogical approaches to support language learning in diverse contexts. This approach will not
only benefit learners in self-study environments but also provide valuable learning strategies and
resources to those in classroom settings or language labs. By researching the integration of
ChatGPT with other language learning tools, we can uncover the language learning outcomes
across various educational settings.

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 176

Conclusion
This mixed-method research directly addresses several prevailing multiple research gaps
marked by a scarcity of empirical studies on generative AI impact on SDLL in informal settings.
It also reveals key opportunities as well as challenges in the self-directed language learning
process, providing crucial resources for educational institutions, policymakers, and curriculum
developers to make well-informed decisions when considering integrating different generative
AI technologies into language education. The present study suggests that while ChatGPT offers
many advantages for SDLL, the successful integration of such generative AI tools for supporting
language learning practices depends on how the learners leverage their motivation, self-
management, and self-monitoring. As generative AI is increasingly integrated into people’s daily
lives, learners, educators, and researchers should seek ways to maximize the potential of
generative AI tools to cultivate learner autonomy, agency, and engagement in a variety of
manners.

Statement on Ethics
The study was approved by Indiana University Institutional Review Board with ID: 20611.
Informed consent was obtained from all participants, and their privacy rights were strictly
observed. The participants were protected by hiding their personal information during the
research process.

Declaration of Competing Interest

The authors declare no conflicts of interest.

Acknowledgment
We thank the Open Education Research Fellowship for mentorship and support throughout this
study.

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 177

References

Adeshola, I., & Adepoju, A. P. (2023). The opportunities and challenges of ChatGPT in
education. Interactive Learning Environments, 1-14.
[Link]
Al-khresheh, M. (2024). Bridging technology and pedagogy from a global lens: Teachers’
perspectives on integrating ChatGPT in English language teaching. Computers and
Education: Artificial Intelligence, 6, 100218. [Link]
Anggoro, K. J., & Pratiwi, D. I. (2023). Fostering self-assessment in English learning with a
generative AI platform: A case of Quizziz AI. Studies in Self-Access Learning Journal,
14(4), 489–501. [Link]
Baidoo-Anu, D., & Ansah, L. O. (2023). Education in the era of generative artificial intelligence
(AI): Understanding the potential benefits of ChatGPT in promoting teaching and
learning. Journal of AI, 7(1), 52-62. [Link]
Bao, M. (2019). Can home use of speech-enabled artificial intelligence mitigate foreign language
anxiety–investigation of a concept. Arab World English Journal (AWEJ) Special Issue on
CALL, (5). [Link]
Baskara, R. (2023a). Exploring the implications of ChatGPT for language learning in higher
education. Indonesian Journal of English Language Teaching and Applied Linguistics,
7(2), 343-358. [Link]
Baskara, R. (2023b). Unravelling lifelong learning prospects via AI-enhanced instruction: The
role of ChatGPT in English language teaching. Innovation in ELT for Inclusive and
Sustainable Education, 64.
Bin-Hady, W. R. A., Al-Kadi, A., Hazaea, A., & Ali, J. K. M. (2023). Exploring the dimensions
of ChatGPT in English language learning: A global perspective. Library Hi Tech, (ahead-
of-print).
Botero, G. G., Questier, F., & Zhu, C. (2019). Self-directed language learning in a mobile-
assisted, out-of-class context: Do students walk the talk? Computer Assisted Language
Learning, 32(1-2), 71–97. [Link]
Bowen, G. A. (2009). Document analysis as a qualitative research method. Qualitative Research
Journal, 9(2), 27-40. [Link]
Braun, V., & Clarke, V. (2022). Toward good practice in thematic analysis: Avoiding common
problems and be (com)ing a knowing researcher. International Journal of Transgender
Health, 1-6. [Link]

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 178

Brookfield, S. D. (2009). Self-directed learning. In R. Maclean and D. Wilson (Eds.),

International handbook of education for the changing world of work: Bridging academic
and vocational learning (pp. 2615-2627). Springer Netherlands.
Brynjolfsson, E., Li, D., & Raymond, L. R. (2023). Generative AI at work (No. w31161).
National Bureau of Economic Research.
Chan, C. K. Y., & Hu, W. (2023). Students’ voices on generative AI: Perceptions, benefits, and
challenges in higher education. International Journal of Educational Technology in
Higher Education, 20(1), 43. [Link]
Creswell, J., & Plano Clark, V. (2007). Designing and conducting mixed methods research.
Sage.
Firat, M. (2023). How chat GPT can transform autodidactic experiences and open education.
Department of Distance Education, Open Education Faculty, Anadolu University.
[Link]
Fisher, M. J., & King, J. (2010). The self-directed learning readiness scale for nursing education
revisited: A confirmatory factor analysis. Nurse Education Today, 30(1), 44-48.
doi:10.1016/[Link].2009.05.020
Garrison, D. R. (1997). Self-directed learning: Toward a comprehensive model. Adult Education
Quarterly, 48(1), 18–33. [Link]
Gibbons, M. (2002). The self-directed learning handbook: Challenging adolescent students to
excel. John Wiley & Sons.
Godwin-Jones, R. (2019). Riding the digital wilds: Learner autonomy and informal language
learning. Language Learning & Technology, 23(1), 8–25. [Link]
Hwang, G. J., & Chen, N. S. (2023). Editorial position paper: Exploring the potential of
generative artificial intelligence in education: Applications, challenges, and future
research directions. Educational Technology & Society, 26(2).
[Link]
Jeon, J. (2022). Exploring a self-directed interactive app for informal EFL learning: a self-
determination theory perspective. Education and Information Technologies, 27(4), 5767-
5787. [Link]
Jumabekovna, E. A. (2024). The Role of Language Learning Apps in Enhancing Vocabulary of
B2 Level Students. International Journal of Pedagogics, 4(01), 91-95.
[Link]
Lee, Y. J., Davis, R. O., & Lee, S. O. (2024). University students’ perceptions of artificial
intelligence-based tools for English writing courses. Online Journal of Communication
and Media Technologies, 14(1), e202412. [Link]
Knowles, M. S. (1968). Andragogy, not pedagogy. Adult Leadership, 16(10), 350-352, 386.
Online Learning Journal – Volume 28 Issue 3 – September 2024
 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 179

Lin, X. (2023). Exploring the role of ChatGPT as a facilitator for motivating self-directed
learning among adult learners. Adult Learning,
[Link]

Li, B., Bonk, C. J., Wang, C., & Kou, X. (2024). Reconceptualizing self-directed learning in the
era of generative AI: An exploratory analysis of language learning. IEEE Transactions on
Learning Technologies, 17, 1515–1529. [Link]
Li, Z., & Bonk, C. J. (2023). Self-directed language learning with Duolingo in an out-of-class
context. Computer Assisted Language Learning.
[Link]
Liu, B. (2023). Chinese university students’ attitudes and perceptions in learning English using
ChatGPT. International Journal of Education and Humanities, 3(2), 132-140.
[Link]
Liu, G., & Ma, C. (2023). Measuring EFL learners’ use of ChatGPT in informal digital learning
of English based on the technology acceptance model. Innovation in Language Learning
and Teaching, 1–14. [Link]
Mishra, P., Pandey, C. M., Singh, U., Gupta, A., Sahu, C., & Keshri, A. (2019). Descriptive
statistics and normality tests for statistical data. Annals of Cardiac Anaesthesia, 22(1), 67.
[Link]
Mogavi, R. et al. (2024). ChatGPT in education: A blessing or a curse? A qualitative study
exploring early adopters’ utilization and perceptions. Computers in Human Behavior:
Artificial Humans, 2(1), [Link]
Morrison, D., & McCutheon, J. (2019). Empowering older adults’ informal, self-directed
learning: Harnessing the potential of online personal learning networks. Research and
Practice in Technology Enhanced Learning, 14(1), 10. [Link]
019-0104-5
Niyozov, N., Bijanov, A., Ganiyev, S., & Kurbonova, R. (2023). The pedagogical principles and
effectiveness of utilizing ChatGPT for language learning. In E3S Web of
Conferences (Vol. 461, p. 01093). EDP Sciences.

Preiksaitis, C., & Rose, C. (2023). Opportunities, challenges, and future directions of generative
artificial intelligence in medical education: Scoping review. JMIR Medical Education, 9,
[Link]
Rahman, M. M., & Watanobe, Y. (2023). ChatGPT for education and research: Opportunities,
threats, and strategies. Applied Sciences, 13(9), 5783.
[Link]

Online Learning Journal – Volume 28 Issue 3 – September 2024

 Exploring the Utility of ChatGPT for Self-Directed Online Language Learning 180

Rasul, T., Nair, S., Kalendra, D., Robin, M., de Oliveira Santini, F., Ladeira, W. J., Sun, M.,
Day, I., Rather, R. A., & Heathcote, L. (2023). The role of ChatGPT in higher education:
Benefits, challenges, and future research directions. Journal of Applied Learning and
Teaching, 6(1). [Link]
Song, L., & Hill, J. R. (2007). A conceptual model for understanding self-directed learning in
online environments. Journal of Interactive Online Learning, 6(1), 27-42.
Tai, T. Y., & Chen, H. H. J. (2023). The impact of Google Assistant on adolescent EFL learners’
willingness to communicate. Interactive Learning Environments, 31(3), 1485-1502.
[Link]
Yu, H., & Guo, Y. (2023, June). Generative artificial intelligence empowers educational reform:
current status, issues, and prospects. Frontiers in Education, 8,1183162.
[Link]
Williamson, S. N., (2007). Development of a self-rating scale of self-directed learning. Nurse
Researcher, 14(2), 66-83. [Link]
.com/openview/c7980aea8ee20b570c57e9102cf5b9ea/1?pqorigsite=gscholar&c
bl=33100
Zhu, M., & Bonk, C. J. (2019). Designing MOOCs to facilitate participant self-monitoring for
self-directed learning. Online Learning, 23(4), 106–134.
[Link]

Online Learning Journal – Volume 28 Issue 3 – September 2024