The REFLECT model: cultivating beginning language learner agency through eTandem conversations, transcriptions, and reflections

4.1.1 Recognizing gaps and transcending the curriculum

In the early stages, students primarily relied on classroom exposure to prepare for their first eTandem conversation. However, the transcribing process following the first exchange acted as a metacognitive trigger, helping them notice input gaps. This aligns with metacognitive knowledge, specifically learners’ understanding of their own linguistic limitations and the demands of the task. For instance, Xiaolin’s realization that she needed to learn vocabulary not included in class content marks a shift from passive or reactive to proactive behavior:

I had to look up words and pronunciations of words […]. I spent many hours doing that extra research and I think it was worth it (Xiaolin, post-conversation 1)

In the second conversation, Xiaolin highlighted the need for authentic, contextual input for language learning, and began to embrace the unpredictable nature of language use:

I thought I knew some basic words about professions but I neglected that these words only work for a small group of professions. […] I think I can never be fully prepared for this. I can learn as I am exposed to more words. (Xiaolin, post-conversation 2).

This awareness reflects PLLT’s emphasis on agentic engagement and the recognition that meaningful input often lies outside structured environments.

4.1.2 Input-seeking through transcription: from noticing to planning

The transcription process emerged as a powerful space for metacognitive regulation, specifically monitoring and evaluation, and for input-seeking behavior, especially as students confronted what they had misunderstood or not noticed during the exchange. Students identified shortcomings in listening comprehension and used these moments to seek new input. For instance, Feiyang noticed discourse markers while transcribing, reporting in her reflection post-conversation 2:

I think I learned a lot of words when I try to write the transcript. For example, “Du coup”,“plein de choses”, “enfin”.

Similarly, Xin explained “These words confused me […] but actually they are just transition words” (Xin, post-conversation 2).

This awareness, followed by her decision to compile a list of transition words, illustrates metacognitive strategic planning and the development of proactive learning habits. Though there was sometimes a delay between noticing and action, the reflection cycles helped bridge that gap: “I feel that I need to learn some transition words in French.” (Xin, post-conversation 2). However, despite this recognition, Xin observed in her third reflection that she had not fully addressed the issue: “I think I’ll need to learn more about transition words in French rather than saying ‘eh’ or ‘umm’,” suggesting a discrepancy or delay between awareness of a lack of knowledge and actual information or input-seeking behavior. It was only following her third conversation that she made a list of transition words encountered in her transcript. Her fourth conversation demonstrates her deliberate experimentation with this new vocabulary, thus showing intentionality and proactive behavior in her L2 output. Such learner-driven input-seeking behaviors are central to PLLT’s concept of directed motivational currents, where sustained engagement emerges from meaningful goals and learner autonomy (Dörnyei 2019; Papi and Hiver 2024).

4.1.3 Becoming an independent language learner: personalizing input and strategic resource use

As students gained experience through the iterative REFLECT cycles, they began to personalize their preparation, integrating self-selected resources, experimenting, and testing individual intentional strategies. Several participants indeed noted that while they had prepared questions, they often lacked vocabulary and strategies to appropriately continue the conversation following their partner’s response to keep the conversation flowing. Feiyang, for instance, explained how she planned to adapt her future preparation after her first interaction:

I think [what] I could improve the next time is that I can try to reply to her response instead of directly jumping into the next question. I prepare the conversation by proposing questions I would like to ask. But it is not very sufficient. I would try to guess her possible answers and accumulate relevant vocabulary and practice more by talking with myself (post-conversation 1).

Her third reflection illustrates how she implemented her plan:

I try to respond to her answer. I think this time is more like conservation. I gave some actual comments and now we can resonate with each other. For example, I said we have the same taste.

Feiyang’s reflections exemplify metacognitive planning and the PLLT concept of autonomy-supportive regulation. Additionally, she experimented with video-watching strategy by listening with English subtitles, then French, demonstrating a growing system of input-seeking strategy that is structured, sustained, and intrinsically motivated, explaining: “I watch the video three times. The first time, I listen with the English subtitles, the second time, I listen with French subtitles.” (post-conversation 3).

Hanjia, too, moved from frustration to a structured approach, using transcripts to diagnose weaknesses and then adjusting his preparation accordingly, qualifying the transcribing process as “the main learning tool of the experience”, explaining “It’s like making it up after not taking enough advantage of it” (Hanjia, final reflection). His reflections and conversations demonstrate he identified vocabulary gaps and the challenge of understanding questions with different structures from what he had learned, using the transcribing process and input opportunity to then seek information to fix these shortcomings:

Vocabulary is still the major issue. Another thing is to get familiar with multiple expressions for the same meaning. Sometimes I get it when the question is in one format I learnt, once it changes a little bit, I fail to get it.

In the third conversation, Hanjia reported a more structured preparation based on the shortcomings he had identified, showing a clear attempt at addressing these challenges: “I prepared several sample answers to the possible questions and practiced them” (post-conversation 3).

Finally, Xin connected her personal interest in cinema and music as a student majoring in Media and Arts to motivation and language input: “I remind myself of my motivation [to learn French] by watching films and searching information about film festivals” (final reflection). This is a clear example of how her identity emerged and influenced her proactive behavior, supporting research arguing that language learning is enhanced when aligned with learners’ personal values and interests (Dörnyei 2019; Papi and Hiver 2024).

4.2.1 Initiating and shaping interactions: proactive engagement and goals

Students exercised their agency by gradually modifying how they interacted with their partners. Some students asked to meet more frequently than required, or to connect on social media, while others reoriented conversations to conform to the course requirements or to their interests. These choices reflect strategic planning and goal revision which are characteristic of PLLT. For instance, Hanjia shifted his focus from task completion to skill development, acknowledging that concentrating on his performance for the grades led to superficial engagement:

I worried too much about the grades, so it did not go as a communication but like acting. It is meaningless if I do it this way. (Hanjia, post-conversation 1)

His subsequent reflection revealed a shift in his motivation:

Next time […], I should focus more on the communication itself rather than viewing it as a test.

This evolution from performance to learning focus informed Hanjia’s decisions to choose a pass/fail option for the course, and he noted “Without the grading pressure, the talk was really enjoyable”. Through his metacognitive adjustments of shaping his learning environment to reduce the affective filter, Hanjia demonstrates his capacity to regulate himself and his agency, thus transforming his interactions, while also highlighting the complexity of interaction-seeking behaviors Papi and Hiver (2025) discuss: Hanjia, as a proactive agent, distanced himself from a partially limiting utilitarian perspective of the conversation (demonstrating skills for a grade) to instead connect with another student on the other side of the world. Other students, like Feiyang, reported being more explicitly intrinsically motivated from the start of the exchange:

I tend to think that interview [conversation] is more like a meeting and greeting between friends, rather than an assignment as a whole.

These interaction-seeking behaviors thus support the variety, intertwined, and evolving purposes of exchanges which Papi and Hiver (2024, 2025) advance, as students here leveraged the eTandem conversations as language development opportunities while also perceiving them as authentic social interactions, thus contradicting Atkinson’s comment (2025) that PLLT reduces interaction-seeking behaviors to instrumentality.

4.2.2 Language use and negotiation strategies: balancing comprehension with target language commitment

The students shared they often felt a tension between wanting to practice their linguistic skills and wanting to exchange information no matter the language. While some participants or even their partners relied on English as a lingua franca to overcome language barriers, others actively asked their partners to remain in the target language, revealing differing agentic behaviors to surmount struggles. Xin explained she felt emboldened as she grew more comfortable, explicitly asserting her preference to force the conversation in French: “I asked her to respond in French as well.”

Xiaoyi’s agentic behavior was instead demonstrated in her relying on the value of partial English use as an occasional crutch to check their understanding:

[…] there were times when she would explain in English. This is certainly useful because it is much like a second or third review…

Although reliance on English may hinder immersion, it also enabled deeper comprehension and allowed students to occasionally see connections between English and French, and to note that in less controlled interactions codeswitching is often a common strategy they themselves employ on their own campus, thus gaining awareness of the strategies they already use to maintain communication.

4.2.3 Metacognitive awareness through transcription

Reviewing the recordings of their interactions and transcribing the conversations served as a catalyst for self-awareness and self-assessment of communication. This reflective practice through transcribing enabled participants to critically assess their communicative behaviors, including concerning turn-taking dynamics, and tried to adapt accordingly. For instance, Yifei recognized a pattern of unintentional interruptions:

I noticed I unintentionally interrupted [my partner] several times… I definitely need to pay more attention to this and leave more time for [my partner] to finish first.

Her statement indicates a heightened sensitivity to conversational flow and norms – and an intent to improve turn-taking, key to successful intercultural interaction.

Similarly, Xiaolin connected multisensory input to better learning, thus asking her partner to write down words in the chat:

I think it is hard for me to remember if I don’t know how to pronounce what I need to remember.

I think it is important to combine visual memory and audio memory together for a new language.

Participants reported gaining more confidence with this awareness and demonstrates refinement of strategies, even as simple as they may seem, in regulating cognitive processes, thus highlighting a growing agency as learners developed a growing awareness of what worked for them.

Participants gradually developed confidence to seek clarification rather than pretend understanding. For example, Hanjia shifted his posture from a face-saving approach, acknowledging at first “I have to pretend I understand,” to an acceptance of being a beginner, which supported his learning:

Je ne comprends pas. Comment dit-on ‘couscous’ en anglaise?

This time I did not rush through, I gave it more patience and asked the interviewee to explain more.

Likewise, Feiyang’s reflections also demonstrated early awareness and behaviors to support her needs: “I tr[ied] to say ‘Moins vite, s’il te plaît’ before the recording starts” (post-conversation 2). The variety of strategies students used, such as gestures, repetitions, synonyms, reformulations, chat box writing, to stay in the target language allowed them to claim a sense of control over their experience and learning.

Additionally, the diverse ways of asking questions made students acutely aware of the limitations of languaculture courses and that conversations rarely go in a scripted way:

I prepared the keywords of the questions I am going to ask. Then practised several times before the official interview. I think it is enough for preparation. However, when my correspondent said something that I did not expect, I will become nervous. (Yanwei, post-conversation 1).

The awareness of their linguistic limitations often motivated students to identify what they needed to do to improve and to act on their needs:

Vocabulary is still the major issue. Another thing is to get familiar with multiple expressions for the same meaning. Sometimes I get it when the question is in one format I learnt, once it changes a little bit, I fail to get it. [sic] (Hanjia, post-conversation 3)

Hence, students practiced questions with their TAs, requesting “a variety of questions to say the same thing” or “similar questions with different words”.

4.2.4 Awareness of emotions as a driver of metacognition and intercultural development

Students often commented on the cultural aspects of their conversations and on their emotional responses. Embarrassment led students to reflecting on their expectations and biases, linguistic challenges, and prompted adjustments of behavior. Feiyang explained for example:

During the interview, I didn’t get that the parents of my correspondent are divorced in the first place […] I felt very guilty and embarrassed.

Xiaolin shared similar concern:

I was still nervous because I felt like I was not familiar with the expressions related to family. And family is such a sensitive topic so I didn’t want to be offensive. During the interview, many times I had to ask her to repeat several times and I felt bad about it, especially when she had to explain many times that her dad was on arrêt. I was a little embarrassed but she encouraged me with her thumbs up when I finally understood what she was saying. After the interview, I feel like I could do better if the interview is carried out in person and I felt so grateful that she tolerated my repetitive questions. (Xiaolin, post-conversation 2).

These emotional responses, mostly rooted in embarrassment and face from misunderstanding and the potential to cause offense, motivated and incited the participants to prepare differently and to consider sociocultural context differences between China and France, thus enhancing their intercultural communicative competence, a key domain in PLLT.

4.2.5 Positioning oneself as a beginner

Several participants reported feeling motivated and invested when they felt their interlocutors considered them as genuine conversational partners:

I think my correspondent treats me as a real French. […] This makes me think about my real thoughts […] and inspires me to give more ‘high-quality’ answers. (Feiyang)

This positioning motivated some students to raise the authenticity of their responses by increasing their cognitive and linguistic complexity. This behavior highlights the motivational component of proactive learning: by feeling perceived as legitimate speakers, students were more likely to take risks, “spending more time to prepare” or looking for more personal vocabulary in their preparation. Feiyang’s comment also demonstrates an evolution from scripted exchanges to authentic interaction, which is a central goal of eTandem.

Participants demonstrate that their interaction-seeking behaviors in the eTandem took various forms, as students went into these conversations with individual and contextual differences making them not only shape the interactions but also affect what they proactively paid attention to, differences consistent with Papi and Hiver’s (2025) PLLT theory.

4.3.1 Monitoring feedback

Participants’ reflections revealed that students primarily engaged in feedback monitoring such as observing and interpreting feedback cues, rather than feedback inquiry. While students reported occasional corrections from partners, especially related to pronunciation and vocabulary, they rarely initiated feedback requests. Xiaoyi’s reflection on her pronunciation illustrates how she both monitored and asked for explicit information:

[…] in addition to Camille’s help in correcting some of my pronunciations, I also gained a deeper understanding of the liaison and practiced with her. She didn’t know why [some words have liaisons and some not] either but was very patient in explaining it to me. (Xiaoyi, post-conversation 2)

The fear of “incorrect pronunciation,” was frequently mentioned in reflections as a recurrent concern, underscoring the affective dimension of feedback monitoring and a focus on form over function. However, most students, reported that their partners rarely corrected them explicitly. They indeed explained that partners did not seem to pay attention to their pronunciation, grammar, and pragmatics, perceiving this behavior as prioritizing understanding and an uninterrupted conversation flow, or a form of mistake tolerance, as Xiaolin’s reflection exemplifies:

We are still nice to each other and tolerate each other’s mistakes. Even if there are mistakes and mispronunciations, it didn’t really impede our conversation and our curiosity.

Students in the present context appeared to prioritize communicative flow and social rapport over linguistic correction, a behavior perhaps attributed to their beginner positionality or lack of confidence.

4.3.2 Perceptions of partner expertise on feedback

Several students suggested that their partners may not have positioned themselves as legitimate sources of corrective feedback, perhaps because of a lack of legitimacy, capacity, or responsibility. Yifei’s reflection demonstrates how giving feedback also needs a particular type of expertise on the form of feedback-giving: “She may not know how to help me improve French as well as [my coach] does” (final reflection).

Other participants also noticed that corrections often came in indirect forms. For instance, Xin observed that her partner “reconstructed her [Xin’s] question” without directly saying it was “wrong”, but Xin realized that while her initial question was understood, pragmatics were perhaps not reaching her partner’s standards. She noted that often, even when writing to her partner to set up the conversation, she “would think that [she] would ignore my minor mistakes in my writing and understand my messages”, perhaps showing a hierarchy in linguistic areas for focus on feedback-seeking: a desire to receive more feedback in oral than in written output, or from certain people and contexts rather than other. Xin described that she prepared her conversations with her partner by having similar conversations with her sister who had been learning French longer, thus expecting her sister to give direct and explicit feedback without even being prompted:

when I practiced with my sister, […] she will point out mistakes immediately rather than trying to understand. […] I think this works for me better (Xin, post-conversation 2).

Here, Xin’s behavior illustrates a self-directed strategy of intentionally creating feedback-rich environments through the multiplicity of interactions, providing her with iterative cycles for metacognitive control.

4.3.3 Transcriptions and reflections as feedback tools

Unsurprisingly, conversation transcriptions followed with reflections emerged as source of self-evaluation and metacognition to identify and respond to errors, reporting on perceptions of how they learn best, what supports their learning, and how to improve. These activities lead to agentic behaviors such as note-taking of mistakes and their correct forms, as Yifei explained:

the best way for me to improve my English is to learn from previous mistakes. And I think it works again for my French learning. I do memorize those things I forgot during the conversation clearer afterward (post-conversation 3).

the transcriptions are useful. Because it forces me to review the recordings again and again. And I need to hear my mistakes again and again. […] I am the kind of person who learned a lot from my previous mistakes. So transcriptions definitely help me to correct (final reflection).

This retroactive self-assessment supports findings in prior studies showing transcription as an opportunity for feedback (Cowie 2018; Stilwell et al. 2010), but the REFLECT model seems to have served as a catalyst for metacognitive awareness by encouraging students to identify specific steps and resources to address mistakes they might have identified, or questions they might want to ask in class. This finding demonstrates how guided post-conversation reflections incorporating learner-generated transcriptions supported error noticing, as Lynch (2007) and Cowie (2018) found in their own studies.

4.3.4 Motivating role of positive peer-feedback

Occasional praise and positive feedback on L2 performance from partners, even non-specific comments, were reported as motivating and boosting confidence: “I received some praise from the interviewee about my French and that helped motivating me a lot” (Hanjia, post-conversation 2). This type of general feedback was inspiring, encouraging students to use the target language: “She said my French was good and it was so encouraging!” (Xiaolin, post-conversation 1). Although this is not a central aspect in Papi and Hiver’s work, this affective component of positive feedback encouraged students to sustain their efforts and engage in their conversations.

The recordings of all students’ conversations showed frequent meaning negotiation and indirect feedback, though some students did not always recognize subtle corrections from their partners. While most students did not mention seeking explicit feedback from their partners in their reflections nor did they exhibit such direct behaviors in the conversations, they sought feedback on their L2 performance through the iterative cycle of conversation, transcription of conversation, and reflection, using the transcriptions as objects of self-assessment, catching occasional errors they made by listening to themselves or to their partner’s responses, processing the linguistic information, and noting them in their guided reflections.

While Proactive Language Learning Theory posits that effective learners engage in both monitoring and inquiry, this study shows that feedback-seeking in early-stage of eTandem exchanges may be imbalanced with a strong orientation toward observation, reflection, and indirect learning. Students seem to be relying on what they consider to be more “experts” than just speakers to ask for direct and explicit feedback.

Unlike previous studies in which explicit error correction accounts for the majority of error address (Luo and Yang 2022; Tang et al. 2021), participants in this study felt their improvements came from the exchange itself and repeated listening of the conversation, rather than explicit correction, suggesting that feedback does not need to be overt to be effective at this stage of language learning. Students indeed exhibited proactive agentic behaviors through self-initiated review cycles and with peer rehearsals, even in the absence of direct partner corrections. All students reported enhanced listening skills, the development of informal language, and increased awareness of generational or cultural language differences, viewing the eTandem as valuable for language learning, and the cycle of conversations-transcriptions-reflections as allowing feedback in various forms. This difference in explicitness of error correction could be due to students being beginners in both languages, and French students not wanting to correct their partners too much. This finding challenges traditional assumptions about the centrality of corrective feedback, and supports a more nuanced understanding of agentic learning behaviors in peer-led, intercultural environments, most particularly for beginners. This finding also highlights how metacognitive strategies can be cultivated even in informal, decentralized feedback contexts at the beginning level.

4.4.1 From error noticing to identifying learning strategies

Supporting PLLT’s principle that learners take ownership of their language learning, participants here initiated information-seeking in response to noticing errors or shortcomings in their conversations with their partners, be it during the conversation or in the iterative cycle of reflection and transcription. Feiyang’s reflections exemplify the idea of monitoring, identifying difficulties with grammatical gender during her conversation with her partner, and planning to seek support to address this issue with her language coach:

I found I am quite confused about when to use the female form of adjectives and verbform. I will figure this out with the help of my TA. (post-conversation 4)

This behavior highlights Feiyang’s early planning and the use of both course and external resources thanks to her growing metacognition and awareness of her language goals. Earlier, she shared her classification of her “Grammar problems” and how she chose to seek information or not:

I think my problems with grammar are divided into two types. The first one is about the small details, I consider them important, but I won’t go crazy about these details considering my intention [to have a conversational level]. I think a good strategy for me would be trying to memorize them correctly the first time so that I can spend less time fixing them later. (post-conversation 2)

Despite being a beginner, and instead of adopting a reactive behavior to grammatical issues which she might develop throughout her learning and eventually choose to address, Feiyang’s quote illustrates her foresight, perhaps based on her previous language learning experience with English. Here, she demonstrates her nuanced and contextual understanding of what she values in language learning: long-term efficiency over perfectionism, demonstrating self-regulatory behaviors.

4.4.2 Interpersonal connections as catalysts for information-seeking

Information-seeking behaviors were not limited to wanting to fix form-related issues. The growing affective connection with their partners led students to want to dive a bit more into their contexts, transcending their language to learn more about their cultural and geographic contexts through autonomous inquiry. Jiajia shared:

It was so nice to see a group of lycéens who are learning Chinese, and I felt connected with another part of the world.

Over time, she explained that her connection to her partner deepened her connection to France:

I feel more connected to that part of the world. My perception used to be gray, but now it’s lightened up, motivating me to learn more about francophone cultures.

Jiajia’s comments reveal an evolution from language learning as classroom activity to language learning as an interpersonal and global practice fostering intercultural curiosity. Indeed, as the eTandem unfolded, students reported researching their partners’ cultures, demonstrating that information-seeking transcended linguistic dimensions to include broader cultural domains. This behavior often took the form of researching aspects of their partners’ environment, as Xin explained:

She lives in Bordeaux so I prepared some questions related to the city. For example, questions related to vin rouge.

While surface-level and potentially essentializing, these information-seeking behaviors reflect the interaction between curiosity and agency. Similarly, Xiaoyi reported:

Since we were going to talk about our cities, I searched for famous places in Bordeaux… hoping to discuss them with Camille.

These examples of information-seeking behaviors demonstrate how curiosity prompted metacognitive evaluation and subsequent actions, thus leading students to take more control over the topics and depth of their conversations.

Information-seeking also led some students to confront their own stereotypes and raciolinguistic expectations, reflecting a deeper engagement with cultural complexity, as Yanwei exemplifies:

I noticed that it is very interesting that my partner is not a native French, but French is her first language… I thought immigration was only common in America. (post-conversation 1)

Yanwei’s new awareness of immigration and France’s colonial past, although naïve, points to critical reflection, although her reflection was not necessarily accompanied with a particular thirst to learn more. Similarly, Xiaolin recounted her assumption about Moroccan accents:

I thought since my interviewee is Moroccan, he might have a different accent that could be hard to understand. […] However, to my surprise, both his English and French were so clear… (Xiaolin, post-conversation 4)

Later, she recognized her own essentializing biases of the multiculturalism of her partner:

I just automatically thought that she is a typical French person… ignoring the fact that I already know that her family is Turkish.

Although shallow, these comments demonstrate a heightened awareness not only of intercultural knowledge, but also of the students’ need to check their biases and automatisms when entering conversations.

Further, while the core eTandem interactions were structured, most students continued their conversations informally, trying to remain in French or using English as a lingua franca to learn about each other’s cultures and languages. These unscripted exchanges fostered further exploration and intercultural connection. Yizi noted:

We continued our discussions for another hour… we discussed the different hand gestures French and Chinese people use to demonstrate numbers… I had the chance to meet Noémie’s father via Zoom. (post-conversation 1)

We discussed the usage of plastic in China and France… she was curious about my weekend routine. (post-conversation 2)

Yizi’s comments illustrate how the eTandem emerged as a tool for sustained intercultural inquiry, with information-seeking emerging organically in response to real interpersonal curiosity.

In line with PLLT, information-seeking behaviors reported in this study were strategic, metacognitive, and learner-driven, but they were also deeply connected to social and affective aspects as learners developed their relationships with their partners. Students occasionally used the eTandem to clarify linguistic forms, thus contributing to explicit L2 knowledge (Papi and Hiver 2024), but students mostly used the opportunity to learn about France and the environments and backgrounds of their partners, deepening these intercultural interpersonal connections outside of the framed exchange. While not contradicting PLLT, this finding extends the framework by illustrating how social and emotional connections affect curiosity and reflection to shape information-seeking strategies, areas not deeply addressed by Papi and Hiver. Further, this adds empirical data and nuance to Papi and Hiver’s (2024) hypotheses, as information-seeking behaviors during interactions were not exclusively about L2. This bridges metacognition and intercultural competence by showing how information-seeking is not just academic but identity-expanding by confronting biases and automatisms, including at the beginning level.