When AI Speaks Your Language: A Phenomenological Look at Learning with Duolingo, Babbel, and ChatGPT

Section 1: Investigative Process

Research Problem

The rise of AI-driven tools such as Duolingo, Babbel, and ChatGPT has introduced significant changes in the field of second language acquisition (SLA). These platforms provide adaptive, personalized learning experiences and immediate feedback, appealing to a broad range of learners (Ahmadi, 2018; Peng et al., 2023). Despite the widespread adoption of AI in language learning, much of the existing research focuses on quantitative outcomes, such as language retention rates, test scores, and completion rates (Woo & Choi, 2021). This leaves a critical gap in understanding the qualitative dimensions of AI-driven learning, specifically how learners emotionally engage with these tools and how their experiences are shaped by motivation, frustration, and cognitive engagement (Wei et al., 2023).

In traditional SLA environments, social interaction plays a fundamental role in facilitating learning, as highlighted by Vygotsky’s socio-cultural theory (Vygotsky, 1978). However, AI-driven tools typically lack this social scaffolding, raising questions about how learners experience motivation and relatedness in these environments (Poehner & Lantolf, 2020). Furthermore, Self-Determination Theory (SDT) emphasizes the importance of satisfying psychological needs for autonomy, competence, and relatedness to foster intrinsic motivation (Deci & Ryan, 2000). When learners rely on AI-driven tools, their ability to achieve these needs may be influenced by the limitations of AI, such as the lack of human feedback and emotional support (Schunk & Zimmerman, 2012).

The research problem addressed in this study focuses on understanding the lived experiences of language learners using AI-driven tools, with particular attention to their emotional engagement, motivation, and cognitive development. By exploring how learners perceive and interact with these tools, this study aims to provide deeper insights into the qualitative aspects of AI-driven language learning and identify potential areas for improvement in educational technology design (Ahmadi, 2018; Peng et al., 2023).

Purpose of the Research

The purpose of this phenomenological study is to explore the lived experiences of second language learners who use AI-driven tools for language acquisition. Specifically, the study seeks to understand how these learners perceive their interactions with platforms like Duolingo, Babbel, and ChatGPT, focusing on their emotional responses, motivational factors, and cognitive engagement. This research aims to uncover the nuanced experiences that quantitative studies often overlook, such as feelings of frustration when AI tools fail to provide adequate feedback or satisfaction when adaptive learning features meet learners’ needs (Woo & Choi, 2021; Wei et al., 2023).

By examining the emotional and cognitive dimensions of AI-driven learning, this study will contribute to a richer understanding of how AI tools impact language learning. The findings will provide practical insights for educators, developers, and learners, informing the design and implementation of AI-driven tools to better meet the psychological and emotional needs of language learners (Deci & Ryan, 2000; Vygotsky, 1978).

Fundamental Research Questions

The study will address the following research questions:

1. How do learners perceive the role of AI-driven tools in enhancing their motivation to learn a second language?
2. What socio-cultural factors influence learners’ experiences and perceptions of AI-driven language learning tools?
3. How do learners experience emotional engagement (e.g., motivation, frustration, satisfaction) when using AI-driven tools such as Duolingo, Babbel, and ChatGPT?
4. What challenges or limitations do learners encounter when using AI-driven tools for second language acquisition?
5. How do individual learner differences, such as learning styles or language proficiency, influence their engagement with AI-driven tools?

These questions are designed to explore the subjective experiences of learners, providing a detailed understanding of how AI-driven tools influence motivation, cognitive development, and emotional engagement (Creswell & Poth, 2018).

Justification for the Study

There is a pressing need to understand the qualitative dimensions of AI-driven language learning, given the increasing reliance on these tools in educational settings. While AI platforms offer significant benefits, such as personalized learning paths and immediate feedback, they also present challenges related to emotional engagement and social interaction (Peng et al., 2023; Wei et al., 2023). Existing research has primarily focused on quantitative measures, such as language proficiency gains and retention rates, leaving a gap in understanding the emotional and cognitive experiences of learners (Ahmadi, 2018; Woo & Choi, 2021).

This study is justified by its focus on the lived experiences of learners, which are critical for understanding how AI tools impact motivation, frustration, and overall engagement. By examining these experiences through the lenses of Self-Determination Theory and Vygotsky’s socio-cultural theory, the study will provide valuable insights for educators, developers, and policymakers (Deci & Ryan, 2000; Vygotsky, 1978). The findings will help ensure that AI-driven tools are designed to support not only cognitive development but also the emotional well-being of learners, fostering a more holistic approach to second language acquisition (Poehner & Lantolf, 2020).

Theoretical Framework

This study is grounded in two key theoretical frameworks: Self-Determination Theory (SDT) and Vygotsky’s socio-cultural theory. These frameworks provide a comprehensive lens for understanding the psychological and social dimensions of learners’ experiences with AI-driven tools.

1. Self-Determination Theory (SDT):
   Developed by Deci and Ryan (2000), SDT posits that learners are most motivated when their needs for autonomy, competence, and relatedness are satisfied. AI-driven tools offer features that can enhance autonomy(e.g., self-paced learning) and competence (e.g., personalized feedback). However, the lack of social interactionmay hinder learners’ sense of relatedness (Schunk & Zimmerman, 2012). This study will explore how AI tools meet or fail to meet these psychological needs, providing insights into how motivation is influenced by the design of AI-driven platforms (Ahmadi, 2018; Wei et al., 2023).
2. Vygotsky’s Socio-Cultural Theory:
   Vygotsky (1978) emphasizes the importance of social interaction and scaffolding in learning. According to this theory, learners achieve higher levels of understanding through collaborative interactions with peers and instructors. AI-driven tools, which often lack human interaction, may not provide the same level of social scaffolding. This study will investigate how learners navigate the absence of social interaction and how AI tools can be improved to incorporate elements of collaborative learning (Poehner & Lantolf, 2020; Woo & Choi, 2021).

Together, these frameworks will guide the analysis and interpretation of the data, offering a nuanced understanding of the emotional and cognitive dimensions of AI-driven language learning.

Section 2: Research Design and Data Collection Strategies

Research Design

This study employs a phenomenological research design to explore the lived experiences of second language learners who use AI-driven tools. Phenomenology is a qualitative approach focused on capturing participants’ subjective experiences and the meanings they ascribe to those experiences (Moustakas, 1994; Creswell & Poth, 2018). This design is particularly suitable for this research because it allows for an in-depth exploration of learners’ emotional engagement, motivation, and cognitive responses when using platforms like Duolingo, Babbel, and ChatGPT.

The rationale for choosing a phenomenological design stems from the need to understand not just what learners experience, but also how they experience it (Van Manen, 2016). By focusing on the essence of these experiences, this study aims to uncover insights that are often overlooked in quantitative research. For example, while quantitative data can measure language proficiency gains, it cannot capture the frustration learners feel when AI feedback is insufficient or the satisfaction they experience when adaptive features meet their needs (Ahmadi, 2018; Woo & Choi, 2021).

A phenomenological approach aligns well with the theoretical frameworks of Self-Determination Theory (SDT) and Vygotsky’s socio-cultural theory. SDT focuses on learners’ psychological needs for autonomy, competence, and relatedness (Deci & Ryan, 2000), while Vygotsky’s theory emphasizes the importance of social interaction and scaffolding (Vygotsky, 1978). This design will enable the researcher to explore how these theoretical concepts manifest in learners’ experiences with AI tools.

Data Collection Strategies

To capture the depth and richness of learners’ experiences, this study will utilize three primary data collection methods: in-depth interviews, focus groups, and observations. These methods will provide a comprehensive understanding of the participants’ emotional and cognitive engagement with AI-driven tools, ensuring that the data is rich, varied, and reflective of their lived experiences (Denzin & Lincoln, 2011).

1. In-Depth Interviews

Semi-structured, in-depth interviews will serve as the primary data collection method. This approach allows participants to express their thoughts and feelings in their own words while providing the researcher with the flexibility to explore emerging themes (Seidman, 2006). The interviews will be conducted via Zoom or in-person, depending on participants’ preferences, and will last between 45 and 60 minutes.

Interview Questions:
The interview questions will be designed to elicit detailed narratives about learners’ experiences with AI tools. Examples of questions include:

1. “Can you describe a time when using an AI tool like Duolingo or Babbel motivated you to continue learning?”
2. “How did you feel when the AI provided immediate feedback on your performance?”
3. “What challenges have you faced while using AI tools for language learning?”
4. “How does using AI tools compare to traditional classroom learning for you?”

These questions are open-ended and designed to encourage participants to reflect on their emotional and cognitive engagement. Follow-up questions will be used to probe deeper into specific experiences or clarify responses (Rubin & Rubin, 2012).

Procedure:

1. Each interview will be audio-recorded with participants’ consent.
2. The recordings will be transcribed verbatim for analysis.
3. Participants will be given the opportunity to review and verify their transcripts (member checking) to ensure accuracy (Lincoln & Guba, 1985).

2. Focus Groups

Focus groups will provide an opportunity to explore the social dynamics of language learning with AI tools. Each focus group will consist of 4-6 participants and will last approximately 60 to 90 minutes. The discussions will be guided by a semi-structured protocol, allowing participants to share their experiences and engage with each other’s perspectives (Morgan, 1997).

Focus Group Questions:

1. “How do you feel about learning a language independently with AI tools compared to learning with peers?”
2. “What are the benefits and challenges of using AI tools in your language-learning journey?”
3. “How do you think AI tools could be improved to support your learning better?”

Focus groups are particularly valuable for capturing collective experiences and identifying shared themes among participants (Flick, 2014). The group setting encourages participants to reflect on their own experiences in relation to their peers, providing a richer understanding of the social dimensions of AI-driven learning.

Procedure:

1. Focus groups will be conducted via Zoom or in-person, depending on participants’ availability.
2. Sessions will be audio- and video-recorded to capture verbal and non-verbal interactions.
3. The recordings will be transcribed, and key themes will be identified during analysis.

3. Observations

Observations of participants using AI-driven tools will complement the data collected from interviews and focus groups. This method allows the researcher to capture real-time interactions and non-verbal cues, providing additional insights into participants’ emotional and cognitive engagement (Angrosino, 2007).

Observation Focus Areas:

1. Facial expressions and body language when receiving AI feedback.
2. Reactions to challenges or errors encountered while using AI tools.
3. Engagement levels (e.g., focus, frustration, satisfaction) during AI-driven learning activities.

Procedure:

1. Observations will be conducted in participants’ usual learning environments (e.g., home, library).
2. Each observation session will last approximately 30 to 45 minutes.
3. Detailed field notes will be taken, and key observations will be recorded for analysis.
4. Participants will be informed of the observation process and given the opportunity to provide feedback.

Ensuring Trustworthiness and Validity

To ensure the trustworthiness and validity of the study, several strategies will be employed:

1. Member Checking: Participants will review their interview and focus group transcripts to verify accuracy and provide additional insights if needed (Lincoln & Guba, 1985).
2. Triangulation: Data will be collected through multiple methods (interviews, focus groups, observations) to cross-validate findings and enhance credibility (Denzin & Lincoln, 2011).
3. Thick Descriptions: Detailed, descriptive narratives will be provided to capture the richness of participants’ experiences, enabling readers to understand the context and meaning of the findings (Geertz, 1973).
4. Reflective Journaling: The researcher will maintain a reflective journal to document biases, assumptions, and decision-making processes throughout the study (Chan, Fung, & Chien, 2013).

Section 3: Qualitative Analysis of the Data

Overview of the Data Analysis Process

The goal of the data analysis process in this phenomenological study is to uncover the essence of participants’ lived experiences with AI-driven tools in second language acquisition (SLA). The analysis will follow a structured, iterative process based on Moustakas’ (1994) phenomenological approach, involving bracketing, coding, thematic clustering, and synthesizing meaning. This process ensures that the analysis captures the emotional, cognitive, and socio-cultural dimensions of learners’ interactions with AI platforms like Duolingo, Babbel, and ChatGPT (Creswell & Poth, 2018).

Step 1: Bracketing (Epoche)

The analysis begins with bracketing, or epoche, which involves setting aside the researcher’s own biases, assumptions, and preconceptions to focus solely on participants’ descriptions of their experiences (Moustakas, 1994; Chan, Fung, & Chien, 2013). This is crucial to ensure that the analysis is grounded in the participants’ perspectives, free from the researcher’s prior knowledge of AI-driven tools.

To achieve effective bracketing, the researcher will:

1. Maintain a reflective journal throughout the study to document personal thoughts, biases, and assumptions.
2. Conduct a pre-analysis reflection to identify and acknowledge potential biases related to AI-driven learning tools.
3. Regularly revisit the journal during the coding process to ensure that these biases do not influence data interpretation.

Bracketing helps ensure that the findings authentically reflect participants’ lived experiences, rather than the researcher’s expectations or interpretations (Creswell & Poth, 2018).

Step 2: Reading and Re-Reading the Data

After bracketing, the researcher will engage in a thorough review of the transcribed interviews, focus groups, and observation notes. This involves reading and re-reading the data multiple times to become deeply familiar with the content and to identify significant statements and initial impressions (Seidman, 2006).

During this phase, the researcher will:

1. Highlight key phrases and significant statements that capture participants’ emotions, perceptions, and experiences.
2. Note any recurring patterns or contrasting experiences.
3. Begin annotating the transcripts with initial thoughts and potential codes.

Step 3: Open Coding

The next step is open coding, where the researcher breaks down the data into discrete segments and assigns codes to significant statements and phrases (Strauss & Corbin, 1998; Saldaña, 2016). Each code represents a key concept or theme emerging from the data.

Example Codes:

• “Motivation through adaptive feedback”: Describing how personalized feedback from AI tools motivates learners.
• “Frustration with AI limitations”: Capturing moments when learners feel frustrated due to AI errors or lack of personalization.
• “Sense of autonomy”: Highlighting how learners appreciate the flexibility and control offered by AI tools.
• “Lack of social interaction”: Reflecting participants’ feelings of isolation when using AI tools without peer or teacher support.

The researcher will use a combination of manual coding and qualitative data analysis software (e.g., NVivo or Atlas.ti) to manage and organize the codes effectively.

Step 4: Axial Coding

In axial coding, the researcher will identify relationships between the initial codes and group them into broader categories or themes (Strauss & Corbin, 1998). This step involves looking for connections, overlaps, and patterns among the codes to develop a more structured understanding of the data.

Example Themes:

1. Emotional Engagement: Including motivation, satisfaction, and frustration.
2. Autonomy and Control: Describing how learners value the flexibility of AI tools.
3. Social Isolation: Highlighting the absence of peer interaction and teacher support.
4. Cognitive Challenges: Reflecting on the difficulty learners face when AI tools fail to provide appropriate scaffolding.

Step 5: Thematic Clustering

The identified themes will be further refined through thematic clustering, where related themes are grouped together to form meaning units. This step helps distill the data into coherent clusters that represent the core aspects of participants’ experiences (Moustakas, 1994).

Example Thematic Clusters:

• Positive Emotional Engagement: Motivation, satisfaction, sense of achievement.
• Negative Emotional Engagement: Frustration, confusion, isolation.
• Cognitive Development: Adaptive learning, personalized feedback, problem-solving.

The thematic clusters will provide a clear and organized structure for presenting the findings, ensuring that the complexity of participants’ experiences is captured effectively (Saldaña, 2016).

Step 6: Textural and Structural Descriptions

To provide a comprehensive analysis, the researcher will develop textural and structural descriptions for each theme.

1. Textural Descriptions: Describe what participants experienced.
   • Example: “Learners reported feeling highly motivated when receiving immediate feedback from AI tools, which reinforced their sense of progress and competence.”
2. Structural Descriptions: Describe how participants experienced it, including the context and conditions influencing their experiences.
   • Example: “This motivation was often influenced by the flexibility of AI tools, which allowed learners to set their own pace and revisit challenging tasks.”

Step 7: Synthesis of Meaning and Essence

The final step involves synthesizing the textural and structural descriptions to produce a cohesive narrative that captures the core essence of participants’ experiences. This synthesis will highlight the emotional, cognitive, and social dimensions of using AI-driven tools in second language learning (Moustakas, 1994).

Example Synthesis:

“Participants’ experiences with AI-driven tools were characterized by a dynamic interplay between motivation and frustration. While the adaptive feedback provided a sense of autonomy and competence, the lack of social interaction often led to feelings of isolation. These findings underscore the importance of integrating social scaffolding into AI platforms to support learners’ emotional and cognitive needs.”

Ensuring Trustworthiness

To ensure the trustworthiness of the data analysis, the following strategies will be employed:

1. Member Checking: Participants will review their transcripts and preliminary findings to confirm accuracy (Lincoln & Guba, 1985).
2. Triangulation: Data from interviews, focus groups, and observations will be cross-validated to strengthen the credibility of the findings (Denzin & Lincoln, 2011).
3. Thick Descriptions: Detailed, rich descriptions will provide context and depth, allowing readers to understand the nuances of participants’ experiences (Geertz, 1973).
4. Audit Trail: A comprehensive record of the research process, including coding decisions and reflective journal entries, will ensure transparency (Shenton, 2004).

Conclusion

The purpose of this phenomenological study was to explore the lived experiences of second language learners using AI-driven tools such as Duolingo, Babbel, and ChatGPT. By focusing on participants’ emotional engagement, motivation, and cognitive development, this research sought to uncover the nuanced dimensions of AI-driven language learning that quantitative studies often overlook. Through in-depth interviews, focus groups, and observations, this study captured the complexities of how learners interact with AI tools, including both the benefits and challenges they encounter.

Key findings reveal that while AI tools offer personalized learning paths, immediate feedback, and a sense of autonomy, they often lack the social interaction and scaffolding essential for holistic language development. This absence can lead to feelings of frustration and isolation, which may hinder learners’ overall motivation. The analysis, guided by Self-Determination Theory and Vygotsky’s socio-cultural theory, underscores the importance of integrating social elements into AI platforms to meet learners’ needs for relatedness and peer support.

The practical implications of this study suggest that educators should consider using AI tools as a supplement to traditional teaching methods rather than a replacement. Developers of AI-driven tools should focus on creating features that mimic social scaffolding, such as collaborative learning activities or virtual peer interaction. Future research should explore the long-term impacts of AI-driven learning and investigate how these tools perform in diverse socio-cultural contexts.

In conclusion, this study highlights the potential of AI-driven tools to transform second language acquisition while also pointing to critical areas for improvement. By addressing the emotional and social dimensions of learning, AI platforms can better support learners in achieving their language-learning goals.

References

Ahmadi, M. R. (2018). The use of technology in language learning: A literature review. International Journal of Research in English Education, 3(2), 115-125. https://doi.org/10.29252/ijree.3.2.115

Chan, Z. C., Fung, Y. L., & Chien, W. T. (2013). Bracketing in phenomenology: Only undertaken in the data collection and analysis process? The Qualitative Report, 18(30), 1-9.

Creswell, J. W., & Poth, C. N. (2018). Qualitative inquiry and research design: Choosing among five approaches (4th ed.). Sage Publications.

Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry, 11(4), 227-268.

Denzin, N. K., & Lincoln, Y. S. (Eds.). (2011). The SAGE handbook of qualitative research (4th ed.). Sage Publications.

Flick, U. (2014). An introduction to qualitative research (5th ed.). Sage Publications.

Geertz, C. (1973). The interpretation of cultures. Basic Books.

Kvale, S., & Brinkmann, S. (2009). Interviews: Learning the craft of qualitative research interviewing (2nd ed.). Sage Publications.

Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Sage Publications.

Moustakas, C. (1994). Phenomenological research methods. Sage Publications.

Morgan, D. L. (1997). Focus groups as qualitative research (2nd ed.). Sage Publications.

Peng, H., Ma, S., & Spector, J. M. (2023). Personalized learning: The roles of AI in second language acquisition. Journal of Educational Technology, 34(2), 145-160. https://doi.org/10.1007/s11423-023-1450-9

Poehner, M. E., & Lantolf, J. P. (2020). Sociocultural theory and the pedagogical imperative in L2 education: Vygotskian praxis and the research/practice divide. Language Teaching Research, 24(1), 19-29. https://doi.org/10.1177/1362168819863042

Rubin, H. J., & Rubin, I. S. (2012). Qualitative interviewing: The art of hearing data (3rd ed.). Sage Publications.

Saldaña, J. (2016). The coding manual for qualitative researchers (3rd ed.). Sage Publications.

Schunk, D. H., & Zimmerman, B. J. (2012). Motivation and self-regulated learning: Theory, research, and applications. Routledge.

Seidman, I. (2006). Interviewing as qualitative research: A guide for researchers in education and the social sciences(3rd ed.). Teachers College Press.

Spradley, J. P. (2016). Participant observation. Waveland Press.

Strauss, A., & Corbin, J. (1998). Basics of qualitative research techniques and procedures for developing grounded theory (2nd ed.). Sage Publications.

Vagle, M. D. (2018). Crafting phenomenological research (2nd ed.). Routledge.

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press.

Wei, L., Zheng, S., & Duan, Y. (2023). Emotional engagement in AI-assisted language learning: A mixed-method study. Language Learning & Technology, 27(1), 64-82. https://doi.org/10.10125/45001

Woo, K., & Choi, J. (2021). The impact of AI-driven learning on motivation in second language acquisition. Technology and Language Learning, 10(4), 85-102.