INTRODUCTION
One of the undisputed assumptions about the effects of studying abroad is the extensive access it grants to the kind of quality input and interaction necessary for successful language acquisition. But how much do study abroad learners actually engage in the types of interactions known to promote successful language learning? Despite the expectations students or practitioners might have about access to quality interactions, Collentine (Reference Collentine, Long and Doughty2009) stated that “there has been no attempt independently to document in a fully quantified manner the types of input and interaction that learners have abroad” (p. 226). Furthermore, little is known about the access study abroad students have to corrective feedback, or the corrections learners receive on their language production, while interacting in naturalistic settings.
Much of what we know about the links between interaction, corrective feedback, and language development has been found through research either in the classroom or the laboratory in the home country of the learners. This large body of work spans three decades (for meta-analyses, see Li, Reference Li2010; Lyster & Saito, Reference Lyster and Saito2010; Mackey & Goo, Reference Mackey, Goo and Mackey2007) and has continuously affirmed the connection between corrective feedback and successful acquisition. While previous work traditionally explored the role of the teacher or native speaker (NS) as the provider of the feedback, the provision of feedback by language-learning peers has also been explored (e.g., Fujii, Ziegler & Mackey, Reference Fujii, Ziegler, Mackey, Sato and Ballinger2016, Sato & Lyster, Reference Sato, Lyster and Mackey2007, Reference Sato and Lyster2012; Varonis & Gass, Reference Varonis and Gass1985) in addition to studies examining how peers collaborate on tasks (Swain, Reference Swain1997) and engage in language-related episodes (LREs; Swain, Reference Swain, Doughty and Williams1998; Swain & Lapkin, Reference Swain and Lapkin1995, Reference Swain and Lapkin1998). However, these studies typically examine tightly controlled laboratory interactions or task-based classroom interactions. Few studies have examined interactions in naturalistic contexts abroad (but see Fernández García & Martínez Arbelaiz, Reference Fernández García, Martínez Arbelaiz, Boers, Darquennes and Temmerman2007; Fernández-García & Martínez-Arbelaiz, Reference Fernández-García and Martínez-Arbelaiz2014; McMeekin, Reference McMeekin, DuFon and Churchill2006). By employing an interactionist perspective on second language acquisition (SLA; Gass, Reference Gass1997; Gass & Mackey, Reference Gass, Mackey, VanPatten and Williams2015a; Long, Reference Long, Ritchie and Bhatia1996, Reference Long2007; Mackey, Reference Mackey1999), the current study attempts to fill this gap by providing primary data on the interactions learners engage in abroad and how the type of feedback they give and receive changes over the course of a short-term stay, the most popular study abroad option for U.S. students (Grey, Cox, Serafini, & Sanz, Reference Grey, Cox, Serafini and Sanz2015). The study additionally investigates the role of the learners’ first language (L1) in the opportunities to engage with corrective feedback.
BACKGROUND
Interaction and L2 Development
Interaction between language learners and more proficient speakers is one of the driving forces of successful language acquisition. The interaction approach (Gass, Reference Gass1997; Gass & Mackey, Reference Gass, Mackey, VanPatten and Williams2015a; Long, Reference Long, Ritchie and Bhatia1996, Reference Long2007; Mackey, Reference Mackey1999) posits that the combination of exposure to modified input and output and negotiation for meaning via the provision of negative corrective feedback is essential to second language (L2) development. This link between interaction and language development has been empirically tested in more than 100 studies since the 1980s (Keck, Iberri-Shea, Tracy-Ventura, & Wa-Mbaleka, Reference Keck, Iberri-Shea, Tracy-Ventura, Wa-Mbaleka, Ortega and Norris2006; Mackey & Goo, Reference Mackey, Goo and Mackey2007; Russell & Spada, Reference Russell, Spada, Ortega and Norris2006), providing robust evidence on a variety of grammatical and discourse features. Input, the first key component of L2 development, can be modified during interaction to suit the needs of the learner at a given point in time (Mackey, Reference Mackey2012). This most often occurs through negotiation for meaning, adjustments made during conversation that make input more comprehensible to learners. Negotiation for meaning often entails the provision of corrective feedback, which can provide the positive or negative evidence that learners need to “notice the gap” (Schmidt, Reference Schmidt1990) between their own production and that of native or proficient speakers.
SLA researchers have extensively investigated the effects of corrective feedback on various aspects of SLA (Mackey, Reference Mackey2012) both in classroom and laboratory settings. Feedback ranges from simply an indication than an error has occurred and has caused a communication breakdown, such as output-prompting clarification requests (“Sorry, what did you say?”), to metalinguistic explanations of how the error should be corrected (“You need to use the past tense ending –ed. . .”). In this regard, corrective feedback moves are frequently discussed and studied in terms of their relative explicit or implicitness. On the implicit end, recasts occur when an interlocutor reformulates all or part of a student's utterance, a type of feedback that has received considerable attention in previous literature (e.g., Ammar & Spada, Reference Ammar and Spada2006; Carpenter, Jeon, MacGregor & Mackey, Reference Carpenter, Jeon, MacGregor and Mackey2006; Lyster & Ranta, Reference Lyster and Ranta1997; Mackey, Gass, & McDonough, Reference Mackey, Gass and McDonough2000; Mackey & Philp, Reference Mackey and Philp1998; McDonough & Mackey, Reference McDonough and Mackey2006). Explicit forms include metalinguistic or grammatical explanations of the error or providing translations.
The degree to which explicit or implicit feedback is more or less effective has been hotly debated in previous literature (Goo & Mackey, Reference Goo and Mackey2013; Lyster & Saito, Reference Lyster and Saito2010), yet these issues have mainly been discussed in terms of research in language classrooms or laboratory settings. Whether or not implicit or more explicit forms of corrective feedback are the most effective, recasts are the feedback move most often used by teachers in language classrooms (Brown, Reference Brown2016). It has been argued that this is because implicit feedback does not interrupt the flow of meaning making in the classroom and is more naturalistic (Long, Reference Long, Ritchie and Bhatia1996, Reference Long2007). Furthermore, the opportunity learners have to modify their output following correction has been shown to enable learners to compare their productions to their perceptions of the instructor's model, thereby fostering automaticity (Swain, Reference Swain and Hinkel2005). While this is the case for classroom and laboratory settings, corrective feedback—especially corrective feedback provided by peer interlocutors—remains understudied in naturalistic settings.
Peer Interaction and L2 Development
In many language-learning environments, learners are grouped together to work collaboratively on a given task. To better understand how pairs or groups of language learners working together impacts L2 performance and development, researchers have examined the ways in which peers provide each other input, prompt for modified output, and offer corrective feedback. Often, learners briefly transition from a meaning-focused interaction to attend to linguistic form. The nature of these interactions has been operationalized as collaborative dialogue (Swain Reference Swain1997), collaborative scaffolding (Donato, Reference Donato, Lantolf and Appel1994), and LREs (Swain & Lapkin, Reference Swain and Lapkin1998). Previous research demonstrated a connection between peer-to-peer collaboration and language learning (Swain, Brooks, & Tocalli-Beller, Reference Swain, Brooks and Tocalli-Beller2002).
While fewer studies have explored peer-to-peer feedback than those that have examined teacher or NS feedback, some patterns have emerged (for an overview, see Sato, Reference Sato, Nassaji and Kartchava2017). A study by Sippel and Jackson (Reference Sippel and Jackson2015) compared three groups of intermediate learners of German: One group received oral corrective feedback from their course instructor, one group was trained to provide peer feedback to one another, and one was the control group. While both experimental conditions demonstrated significant improvements in terms of grammatical accuracy in the two forms under study, the peer feedback group showed the greatest improvements on posttests and delayed-posttests. Sippel and Jackson argued that when peers provide feedback to each other, they stand to gain not only from receiving the feedback, but also from providing the feedback to their peers. Both receiving and providing oral feedback worked to heighten learners’ awareness of linguistic forms. Various other studies have found that learners tend to negotiate for meaning more often with each other as compared with NSs (e.g., Mackey, Oliver, & Leeman, Reference Mackey, Oliver and Leeman2003; Varonis & Gass, Reference Varonis and Gass1985). Others found that peers provide similar interactive moves as NSs, including modified input, opportunities for modified output, and corrective feedback (Mackey et al., Reference Mackey, Oliver and Leeman2003; Pica, Lincoln-Porter, Paninos, & Linnell, Reference Pica, Lincoln-Porter, Paninos and Linnell1996; Sato & Lyster, Reference Sato, Lyster and Mackey2007). In a study by Mackey et al. (Reference Mackey, Oliver and Leeman2003), adult learner-learner dyads completing a task produced more opportunities for modified output, and child learner-learner dyads produced more modified output than they did with NSs. Additionally, several studies have found that learners tend to self-correct more often when interacting with peers than with NSs (Sato, Reference Sato2007; Shehadeh, Reference Shehadeh2001).
Despite these positive findings, some additional studies have uncovered drawbacks to peer interaction. Loschky & Bley-Vroman (Reference Loschky, Bley-Vroman, Crookes and Gass1993) found that, depending on the task, learners might avoid negotiating for meaning and instead focus exclusively on task completion. Sato and Lyster (Reference Sato and Lyster2012) argued that a common type of feedback provided by peers consists of simple segmentations of their peer-interlocutor's erroneous utterances, and this feedback is of lesser quality than feedback provided by NSs. Other research has found that learners might not attune to feedback provided by peers because they do not always believe in each other's linguistic capabilities (Yoshida, Reference Yoshida2008) or because they simply prefer feedback from teachers over feedback delivered by peers (Chu, Reference Chu2013). Additionally, cultural backgrounds and the avoidance of face-threatening behaviors may affect the degree to which learners are willing to provide each other feedback (Fujii & Mackey, Reference Fujii and Mackey2009; Sato & Lyster, Reference Sato, Lyster and Mackey2007). However, many of the disadvantages for peer-to-peer interaction and feedback are heavily influenced by task conditions, either in the classroom or laboratory, and are typically contextualized in foreign language, nonimmersion settings.
Of the studies that have examined learners interacting and negotiating for meaning together in groups, few have studied learners in naturalistic (nonclassroom, nonlaboratory testing) settings (but see McDonough & Hernández González, Reference McDonough, Hernández González, McDonough and Mackey2013; Polio, Gass, & Chapin, Reference Polio, Gass and Chapin2006; Ranta & Meckelborg, Reference Ranta and Meckelborg2013; Ziegler et al., Reference Ziegler, Seas, Ammons, Lake, Hamrick, Rebuschat, McDonough and Mackey2013). While very early work on L2 interaction did focus on learners interacting in informal conversations (e.g., Long, Reference Long1981), also known as “conversations-for-learning” (for a review, see Kasper & Kim, Reference Kasper, Kim and Markee2015), since then, most studies using an interactionist approach have examined learners in instructional or laboratory settings. The current study examined small groups of peer language learners interacting with one or two NSs (also peers of a similar age range). One similar study (Ziegler et al., Reference Ziegler, Seas, Ammons, Lake, Hamrick, Rebuschat, McDonough and Mackey2013) examined the learning opportunities of German learners at a U.S. institution interacting in weekly conversation groups. This study took a discourse analytic approach to understanding the learners’ development of German conversational style and found that the naturalistic setting of the conversation groups played an important role in the development of native-like conversational styles.
While this is evidence that naturalistic learning is effective for conversational style in foreign language setting, these findings should be extended to learners interacting in conversation groups while studying abroad. Investigating the unique context of studying abroad is critical, especially in light of some evidence that learners abroad do not uniformly take advantage of the opportunities they have to interact in natural settings (e.g., Ranta & Meckelborg, Reference Ranta and Meckelborg2013).
Feedback in Study Abroad
Studying abroad is commonly regarded by students and teachers as one of the best methods of increasing access to quality linguistic input and authentic interaction (Freed, Reference Freed1998). However, empirical research into the effects of study abroad experiences on L2 development has shown mixed results, with some studies indicating limited improvements when compared with learners who stayed at home, especially in short-term programs (e.g., Churchill & DuFon, Reference Churchill, DuFon, DuFon and Churchill2006) and other studies showing domain-specific gains, with study abroad students improving fluency and pragmatic abilities rather than grammatical aspects (e.g., Collentine, Reference Collentine, Long and Doughty2009). Other studies have posited a connection between length of stay abroad and L2 development (e.g., Davidson, Reference Davidson2010), with some results indicating that even small increases in the duration of stay, such as 4 weeks instead of 3, can play a role in subsequent linguistic gains (Llanes & Muñoz, Reference Llanes and Muñoz2009). However, short-term programs (less than 8 weeks), such as the one investigated in the current study, remain the least investigated in the literature (Llanes, Reference Llanes2011). Despite the inconclusive evidence on the effects of study abroad, there is little research on the nature of the interactions students engage in with NSs or peers in the nonformal instructional settings they encounter while abroad (Wilkinson, Reference Wilkinson2002), which leads to the question: Are the interactions that learners are exposed to in study abroad supportive of language acquisition?
A limited number of studies have sought to describe and understand conversations students have with NSs “in the wild” (Hutchins, Reference Hutchins1995) of the host country (Fernández García & Martínez Arbelaiz, Reference Fernández García, Martínez Arbelaiz, Boers, Darquennes and Temmerman2007; Fernández-García & Martínez-Arbelaiz, Reference Fernández-García and Martínez-Arbelaiz2014; McMeekin, Reference McMeekin, DuFon and Churchill2006; Wilkinson, Reference Wilkinson2002). Some have adopted an interactionist perspective (Fernández García & Martínez Arbelaiz, Reference Fernández García, Martínez Arbelaiz, Boers, Darquennes and Temmerman2007; Fernández-García & Martínez-Arbelaiz, Reference Fernández-García and Martínez-Arbelaiz2014; for an overview, see Bryfonski & Mackey, in press) while others have utilized conversational analysis methods (Wilkinson, Reference Wilkinson2002) or qualitative case studies (McMeekin, Reference McMeekin, DuFon and Churchill2006). In two studies of NS learner dyads interacting during informal conversation groups abroad in Spain, Fernández García & Martínez Arbelaiz (Reference Fernández García, Martínez Arbelaiz, Boers, Darquennes and Temmerman2007; Fernández-García & Martínez-Arbelaiz, Reference Fernández-García and Martínez-Arbelaiz2014) descriptively analyzed how learners negotiated for meaning and whether or not learners had opportunities for feedback and uptake. These researchers found that recasts were the most frequent moves during these interactions, followed by other moves that the researchers termed “lexical assistance,” “form assistance,” and “completion.” The learners in this study often self-initiated LREs by prompting their interlocutors for input.
Other researchers have also demonstrated how learners adopt classroom roles and discourse structures while interacting abroad. In observations and interviews with learners studying abroad in France, Wilkinson (Reference Wilkinson2002) found that learners interacting with their host families and other interlocutors, such as travel agents and children, relied mostly on classroom-instructional norms—with the NSs adopting the role of the teacher and the learners taking on the role of the student. Wilkinson concluded that “perhaps immersion in a target-language community during study abroad does not always take students as far beyond the classroom as one might intuitively believe” (p. 169). Similarly, McMeekin (Reference McMeekin, DuFon and Churchill2006) reported on a case study of five students interacting while studying abroad in Japan in their classroom and with their host families. McMeekin found that host-family negotiations provided more comprehensible input than classroom discourse, but not output because the NSs corrected or rephrased the learners’ potentially troubled utterances and left no opportunities for modified output. The classroom abroad, however, provided more opportunities for modified output, but focus on form was part of both settings. McMeekin concluded that a combination of both settings abroad would provide students with maximum opportunities for exposures to the types of interaction deemed necessary for language acquisition. However, these studies were cross-sectional; few have looked at the effects of negotiation for meaning and corrective feedback on interlanguage development over time. One longitudinal study by Ranta and Meckelborg (Reference Ranta and Meckelborg2013) examined the quantity and quality of exposure to English that Chinese graduate students had while studying at a Canadian university. Over the course of 6 months, learners used computerized logs to record their daily interactions. Findings indicated that while learners were exposed to high levels of English input, they interacted in English relatively infrequently. These students were not able to engage in the kinds of interactions known to be beneficial to the SLA process (Long, Reference Long, Ritchie and Bhatia1996) despite their immersive setting. Ranta and Meckelborg considered as possible factors that contributed to their findings the learners’ low willingness to communicate, cultural background, and motivation to stay connected with the Chinese expatriate community by using their L1. However, the use of the students’ L1 has not yet been factored into the opportunities learners have to receive input and corrective feedback or to produce modified output while studying abroad.
First Language (L1) Use
The use of the L1 in the context of classroom language learning has been examined and debated in a variety of past work (Cook, Reference Cook2001; Duff & Polio, Reference Duff and Polio1990; Levine, Reference Levine2003; Polio & Duff, Reference Polio and Duff1994), with researchers proposing various situations where the L1 is appropriate to use (or not use) by both teachers and language learners. A study by Antón and DiCamilla (Reference Antón and DiCamilla1999) found that learners working together on a task used their mutual L1 to build collaborative dialogue that enhanced their language acquisition. Antón and DiCamilla rejected the notion that the L1 should be banned from language classrooms and instead advocated for L1 use in certain contexts as a communicative tool. The strategic use of the L1 in foreign language classroom environments has also been advocated for in a variety of previous work (Azkarai & García Mayo, Reference Azkarai and García Mayo2015, Reference Azkarai and García Mayo2017; Tognini & Oliver, Reference Tognini, Oliver, Soler and Jordá2012) with some research showing that despite instructors’ fears (Storch & Wigglesworth, Reference Storch and Wigglesworth2003; Tognini & Oliver, Reference Tognini, Oliver, Soler and Jordá2012), learners do not make excessive use of their L1 when collaborating on tasks (Azkarai & García Mayo, Reference Azkarai and García Mayo2015). A study by Tognini and Oliver (Reference Tognini, Oliver, Soler and Jordá2012) examined how children used their mutual L1 while collaborating on foreign language tasks and found that the learners used their L1 in peer-to-peer interactions to scaffold each other's production and did so more often in negotiations for meaning rather than in form-focused exchanges. Learners might also employ their L1 in self-directed or private-verbal speech when working on a task, which has been shown to aid in their process of reasoning (Centeno-Cortés & Jiménez Jiménez, Reference Centeno-Cortés and Jiménez Jiménez2004).
In immersion contexts, studies have found that L1 use during task-based activities was implemented by students for a variety of reasons, including to manage the task, to focus in on particular vocabulary or grammar items, and to enhance interpersonal interaction, and that it was particularly effective for lower-proficiency students (Swain & Lapkin, Reference Swain and Lapkin2000). Study abroad students are often prohibited from using the L1 in order to take advantage of the immersive environment in which they are learning. Many immersion programs require students to sign a language pledge agreeing to “communicate only in their language of study throughout the program” and touting benefits to the pledge such as “rapid acquisition of linguistic and cultural fluency” (Middlebury Language Schools, 2016). Previous studies have indicated that learners take the pledge very seriously (Dewey, Reference Dewey2004) and some programs penalize students with grade reductions or even suspension from the program for violations (Grey et al., Reference Grey, Cox, Serafini and Sanz2015). One study (Du, Reference Du2013) found positive effects on fluency measures for study abroad students who stated in interviews that they observed the language pledge strictly in comparison to those students who stated that they spoke their L1 whenever they had the opportunity. However, the effectiveness of language pledges is still under investigation, and some authors cite negative effects such as resentment toward the target language (Kuntz & Belnap, Reference Kuntz and Belnap2001) or tension between students who strictly observe the pledge and those who do not (Du, Reference Du2013). To date, little is known about if, how, or why study abroad students who have taken the language pledge use their L1 when interacting with NSs or with each other during their study abroad experiences.
On the basis of the theoretical foundations discussed here and considering current gaps in the literature, the current study will shed light on the following research questions:
RQ1. What is the relationship between the time spent abroad and the amount and type of corrective feedback provided in conversation groups?
RQ2. What differences exist between the corrective feedback provided by native speakers versus nonnative peer interlocutors during study abroad conversation groups?
RQ3. What is the role of the L1 in facilitating opportunities to engage with corrective feedback in these conversation groups?
METHOD
Participants
The study was conducted in the summer of 2016 with 29 participants; 19 were university students from a U.S. University (see Table 1 for an overview of participant biodata). These 19 students were native or very close near-native speakers of English; one student identified as a heritage speaker of Spanish and two other students listed languages other than English as their native languages (Italian and Korean). Other students listed additional L2s at varying levels of proficiency (Hebrew, Russian, French, and Portuguese). Students had been studying Spanish for an average of 8.6 years prior to their trip to Barcelona (range: 4 to 11 years) and were rated intermediate-high by their home institution, meaning they had previously completed two advanced Spanish language courses.
TABLE 1. Participant Biodata
Ten of the NS participants were students at the local university in Barcelona where the study abroad program took place. The study abroad students were required to attend the regularly scheduled conversation groups, and the NS students were paid to act as weekly conversation partners. This program component was meant to develop learner autonomy and to foster the exchange of ideas. Learners and NSs were given no specific instructions or training on how to interact or what to talk about, and they formed groups according to their own preferences. They were required to meet for at least 1 hour twice a week, but in practice, participants spent anywhere between 1 and 4 hours interacting, depending on the day and group.
Context
All study abroad participants were taking part in a faculty-led U.S. institution's 6-week program in Barcelona, Spain, where Catalan is also spoken. Participants in the program were there to continue their study of Spanish, and all their coursework and activities were conducted in Spanish. The program website characterizes its program as emphasizing writing, the development of students’ autonomy as language learners, and the conceptual connections between experiences both within and outside the classroom. While abroad, the students enrolled in three courses on politics, art history, and linguistics, taught in Spanish at the local university by faculty from their home institution and by host faculty. Students were required to complete 12 hours per week of coursework and an additional 10 hours of homework per week. Students also regularly participated in daylong organized excursions with NSs, participated in conversation exchanges, and completed daily reflective writing assignments. Students lived together in a dormitory throughout the course of the program, supervised by an NS instructor.
At the beginning of the program, all students were required to sign a language pledge promising to only speak in Spanish during the duration of the program. Students and faculty were expected to take this pledge very seriously; resident assistants monitored language use in the residence halls and during meals, and they prohibited cell phones during activities. Significant violations to this language pledge could lead to sanctions by the program and home institution.
Data Collection
The current study employed a sequential embedded mixed methods design (Mackey & Bryfonski, in press), with quantitative and qualitative data collected in multiple iterations to inform and follow up on one another (see Figure 1).
FIGURE 1. Research Time Frame and Events.
The main source of data for this study was from audio recordings of conversation exchanges between groups of study abroad students and local NSs (see Table 2 for an overview of the audio-recording data sample). These conversation groups were part of the study abroad program's regular programming. Selected study abroad students also participated in semistructured and stimulated-recall interviews, and they took a custom quiz approximately 4 months after completion of the study abroad program.
TABLE 2. Audio Recordings Overview
Note: All names are pseudonyms. NS = native speaker; NNS = nonnative speaker.
As part of the program's emphasis on student autonomy, students were able to form conversation groups freely with whomever they wished. Some students tended to group together; however, it is not necessarily the case that all students contributed equally to recordings at each time frame sampled. Only four participants appear across all three time intervals.
Students were asked by the study abroad program practitioners to meet in any mutually agreed-upon location, record 30 minutes of natural conversation out of their 1-hour conversations, and send their recordings to the researcher. In practice, the students and NSs met at cafes, bars, and restaurants, or took walks around the city. Collentine and Freed (Reference Collentine and Freed2004) referred to this setting as a “hybrid communicative-learning context” because the students must negotiate both formal learning contexts and purely communicative contexts. In these settings, study abroad learners “often consciously attempt to utilize in communicative interactions explicit knowledge they have attained from their (concurrent classroom) learning experiences” (Collentine & Freed, Reference Collentine and Freed2004, p. 156). This is what Wilkinson (Reference Wilkinson2002) referred to as the “omnipresent classroom” where students and NSs alike adopt classroom roles and discourse patterns, such as requesting or providing feedback, to manage interactions with study abroad students.
Approximately 4 months later, a select number of the study abroad students (n = 9) volunteered to participate in follow-up interviews, back at their U.S. home institution. Interviews followed a semistructured protocol and targeted the following themes of interest: interactions with NSs abroad (questions based on Zappa-Hollman & Duff, Reference Zappa-Hollman and Duff2015), the role of corrections abroad and in the conversation groups, and the use of English abroad (adapted from Duff & Polio, Reference Duff and Polio1990).
Eight of the nine interviewees also took part in a tailor-made quiz following the interview. Previous research has used tailor-made quizzes or tests for learners, following a given treatment to assess recall or the efficacy of collaborative tasks (LaPierre, Reference LaPierre1994; Loewen & Philp, Reference Loewen and Philp2006; McDonough & Sunitham, Reference McDonough and Sunitham2009; Swain & Lapkin, Reference Swain, Lapkin, Bygate, Skehan and Swain2001; Zeng & Takatsuka, Reference Zeng and Takatsuka2009). These tailored quizzes address what learners actually did and said during interactions. The expectation is that when students collaborate, discuss, or receive feedback on a linguistic form, they will be more likely to remember it on the tailored quiz (Loewen & Philp, Reference Loewen and Philp2006). While previous research has found that the majority of LREs that are correctly resolved during collaborative dialogues are retained on tailored posttests (Swain, Reference Swain, Doughty and Williams1998), incorrectly resolved or unresolved LREs lead to much lower rates of accuracy on tailored posttests (Williams, Reference Williams2001).
The tailored quizzes in the current study targeted lexical items and grammatical forms (following McDonough & Sunitham, Reference McDonough and Sunitham2009) that were involved in corrective feedback episodes via either implicit or explicit feedback during the participants’ conversation groups. For example, if a student received corrective feedback or discussed the meaning of the word contraseña in the conversation groups, then on the tailor-made test, they would be given the word password and asked to provide the Spanish word. For grammatical forms, learners were provided with an English phrase or sentence and asked to supply the Spanish translation. For example, if the learner received feedback on the conjugation for a preterit or imperfect verb in a sentence, on the tailored test the learner was presented with an entire sentence in English and asked to say that sentence in Spanish. Each student was tested on two to four words or phrases; 24 words in total were tested. The excerpts were randomly chosen from all the corrective feedback and provision of corrective feedback episodes the student engaged in during data collection. The words were written in English and presented one at a time to the participant. Each participant was asked to say the word or words in Spanish if they could remember them. As Loewen and Philp (Reference Loewen and Philp2006) pointed out: When tailor-made quizzes are used as posttests without pretests, it is impossible to say whether or not the test measures the acquisition of new forms or simply “the consolidation of latent knowledge” (p. 542). Therefore, these quizzes only demonstrate the ability to recall after feedback forms that were previously problematic for students.
Eight interviewees also participated in a stimulated-recall interview after the custom quiz and upon conclusion of the interview. The stimulated-recall interviews were conducted based on the recommendations provided in Gass and Mackey (Reference Gass and Mackey2015b). Previous research has shown that the reactivity and time-delays can cause validity issues as learners move from accessing memories in their short-term memory to retrieving information from their long-term memory, which has been shown to be less reliable (for a description of issues of reactivity, see Egi, Reference Egi and Mackey2007, Reference Egi2008; Leow & Morgan-Short, Reference Leow and Morgan-Short2004). Therefore, as a result of the time delay between the initial conversations and the interviews (4 months), the stimulated-recall aimed to obtain only perceptions and reactions rather than attempt to access processes involved in the interactions. During the stimulated-recall portion, the researcher played between two and seven pre-identified clips from the participant's conversation groups. Each clip involved a corrective feedback move directed toward the participant from either an NS or one of the nonnative-speaker (NNS) peers in the conversation group. Clips in which the participant provided feedback to a peer were also played if they occurred in the data. Each clip was played two times, and the participant was asked to describe what they were thinking at the time or, since these stimulated-recall interviews did not occur immediately following the interactions, simply to describe their reaction to the feedback.
Analysis
To address RQ1 and RQ2, all corrective feedback episodes were identified in the audio recordings of the conversation groups and then transcribed by the researcher.Footnote 1 Means for the number of corrective feedback moves in each conversation group over the three time periods were calculated. The feedback moves were then coded in terms of (a) the interlocutor who produced the feedback (the NS or the NNS study abroad students) and (b) whether the feedback was implicit or explicit in nature. Following Mackey (Reference Mackey2012), feedback was coded with a meaning-based dichotomy. Feedback was coded as implicit feedback when the feedback provider focused exclusively on meaning, that is, by not indicating the error's location, by clarifying understanding in order to move the conversation forward, or by simply negotiating for meaning. Implicit feedback includes recasts (reformulations), clarification requests (prompts), and confirmation checks (see Example 1; NSs appear in bold in all of the examples).
Example 1. Implicit Feedback Coding
Explicit corrections were those that focused on form and offered either specific grammar rules or a translation or where form was discussed explicitly. Metalinguistic feedback or other LREs, where form was explicitly discussed, and translations were both coded as explicit feedback (see Example 2).
Example 2. Explicit Feedback Coding
A trained rater coded a subsample (20%) of the identified feedback moves for both coding features (whether the feedback provider was an NS or NNS and whether the feedback was implicit or explicit), and any differences were discussed for an initial reliability estimate of 85%. The remaining 15% were considered again by the two coders, and a final agreement of 100% was reached.
The responses on the custom quizzes provided additional quantitative data to address RQ1 and RQ2. Each item on the tailored quizzes was graded categorically as either completely correct or incorrect. The total number of correct responses was divided by the total number of items on the tailored quizzes to produce a percentage accuracy score. Accuracy scores were compared with findings from previous studies that utilized tailored quizzes to measure uptake of lexical and grammatical forms following negotiations for meaning (e.g., McDonough & Sunitham, Reference McDonough and Sunitham2009; Williams, Reference Williams2001).
To address RQ3, all instances of L1 English use were identified in the audio recordings of the conversation groups and then were transcribed by the researcher. The use of the L1 was counted in terms c-units, that is, a multiword utterance clause was counted as 1 unit (e.g., I don't know), as was the insertion of a single English word (cómo se dice password?).
We used quantitative analyses to address all three research questions. One-way ANOVAs tested for statistically significant changes in the number of feedback moves across the three time periods, as well as statistical differences between the type of feedback provided (implicit or explicit), the provider of the feedback (NS or NNS), and change in L1 use over time. Effect sizes were calculated by Cohen's d (Cohen, Reference Cohen1988). Qualitative analyses were used to inform and follow up on the quantitative findings for each research question. Semistructured interviews were coded using a thematic, grounded approach using QSR International's qualitative analysis software NVivo. The resulting themes were then associated with relevant excerpts from the audio recordings. The stimulated-recall portion of the interviews were transcribed, and any perceptions or reflections were coded thematically and categorized according to the type of feedback and the feedback provider in each example discussed during the interview.
RESULTS AND DISCUSSION
RQ1: What is the relationship between the time spent abroad and the amount and type of corrective feedback provided in conversation groups?
In total, 176 feedback moves were identified and transcribed; 74 moves at Week 1 (M = 14.8, SD = 4.32), 69 at Week 3 (M = 17.3, SD = 2.06), and 33 at Week 6 (M = 8.25, SD = 5.56). As shown in Figure 2, there was a decrease in the amount of feedback provided by both NSs and peers per conversation group between the first week and the last week of the program.
FIGURE 2. Change in Total Amount of Feedback per Conversation Group Over Time.
Results from a one-way ANOVA demonstrate statistically significant change in total feedback moves over time (F = 4.85, df = 2, p = .03, η2 = .49). An LSD post hoc comparison (as recommended by Howell [Reference Howell2002] and Larson-Hall [Reference Larson-Hall2010] for three comparisons) revealed a statistically significant difference between Weeks 1 and 6 (mean difference = 6.55, 95% CI = .20, 12.9, p = .04), as well as a statistically significant difference between Weeks 3 and 6 (mean difference = 9.00, 95% CI = 2.31, 15.69, p = .01). No difference was found between Weeks 1 and 3 (mean difference = −2.45, 95% CI = −8.80, 3.90, p = .41). Effect sizes revealed a small to medium effect (d = 0.7) for the change in feedback between Weeks 1 and 3, a medium to large effect (d = 1.30) for the change in feedback between Weeks 1 and 6, and large effect (d = 2.10) for the change in feedback between Weeks 3 and 6.
The overall result that the total amount of feedback decreased over time is supported by the interview data. Seven of the interviewees reported that they perceived a change during the course of the 6 weeks of conversation groups. Their comments indicate that they believed the conversations had less of a language focus, were more natural, and led to fewer communication breakdowns by the end of the study abroad program. For example, Cliff (all names pseudonyms) described how he did not feel the need to translate from English to Spanish by the end of the program:
It got to the point . . . that I kind of just started thinking in it. I wasn't thinking like, “How do I say this in Spanish?” I would sort of just say whatever was on my mind.
Jessica also perceived this change and described how the conversation groups shifted from being stilted and awkward to just catching up with Elena, the NS her group met with each week.
It was natural . . . especially towards the end of the sessions because we knew we were going to [Elena] and we knew we could catch up and then it was nice because we all wanted to talk.
In terms of the type of feedback per conversation group, the amounts of implicit and explicit feedback decreased similarly over time (as shown in Figure 3). However, the difference between the total amount of explicit (total = 80) and implicit feedback (total = 96) was not statistically significant.
FIGURE 3. Changes in Type of Feedback per Conversation Group Over Time.
This finding was also supported by the qualitative data. When asked if the NSs in the group corrected the study abroad students all the time, or only when there was a misunderstanding, all nine participants in the interviews agreed that they were not corrected all the time, only when there were misunderstandings. One student said that they were only corrected “when it came up . . . it never felt like an interruption.” Excerpt 1 demonstrates one student receiving implicit feedback on the incorrect use of a lexical item (Note: NSs are highlighted in bold in all of the excerpts.)
Excerpt 1. Implicit Feedback
Maria recalled the word for “translated” correctly on her custom quiz and very clearly remembered this correction during the stimulated-recall interview. She said:
I remembered that one just because when I was saying it I didn't really stutter on it. . . . I was taught this . . . it was something that I was confident in saying . . . then they were like no you're doing it wrong. How many other things have I been saying incorrectly?
However, the success of the implicit feedback in terms of later retrieval of correct forms was mixed. Excerpt 2 is an example of one such implicit correction taken from the first week of the program with the same participant:
Excerpt 2. Implicit Feedback
Maria did not correctly recall the word “trilingüe” on her tailored quiz and stated in the stimulated-recall interview:
I don't think I realized. I don't think I was really sure what I was trying to say. But I just knew “bilingual” and then went with that.
One difference to point out between these two exchanges is that Maria recalled noticing the feedback in Excerpt 1 and modified her output. She also received both an output-prompting clarification request, “They can be?” (line 2), as well as a recast, “Translated” (line 4), adding to the salience of the feedback. However, in Excerpt 2 she received only positive input and did not modify her output, but simply continued on in the conversation, an indication that the feedback the NS provided was most likely not noticed. Or if the feedback was noticed at the time, it only helped her to be sensitive to future patterns in the input, rather than explicit knowledge (for more evidence of salience in implicit feedback, see Leeman, Reference Leeman2003).
On the other hand, several students stated that they explicitly asked the NSs to correct them and often stopped to ask for translations or grammatical explanations. In Excerpt 3, Aubrey asked for explicit feedback on an unknown lexical item from the NS.
Excerpt 3. Explicit Feedback
In Excerpt 3, Audrey did not accurately modify her output. However, the results from the custom quiz indicate that she remembered the correct form. When played this clip during the stimulated-recall interview, Aubrey stated that she remembered the word because she asked about it so many times throughout the program. She said:
There were a few words that came up like “contraseña” that I definitely learned throughout the program. . . . A few, like 20 or so words I had weird contact with. I will always remember them now.
Although Aubrey did not appear to notice the feedback in this particular example during the conversation groups, she successfully learned the word through other encounters in her study abroad (or post–study abroad) experience.
However, explicit correction did not always result in successful recollection. In Excerpt 4 the NSs Diego and Anne negotiated with Cliff to break down his ill-formed utterance. They repeated it together until he understood his error:
Excerpt 4. Explicit Feedback
However, when Cliff was asked to recall how to say “I thought that they were foreigners” on his custom quiz, he was incorrect, although he stated that he considered this kind of feedback helpful. He recalled in the stimulated-recall portion of the interview:
[Diego] would do that it a lot . . . sometimes we would have to ask him to repeat that or sound it out and try to internalize it.
This sentiment was not felt by everyone, however. When Carol reflected on asking for the translation for “spelling” and receiving corrective feedback, she stated “just from her [the NS] telling me, I couldn't remember.” Carol also did not correctly recall the Spanish word for “spelling” during her tailored quiz.
As Mackey et al. (Reference Mackey, Gass and McDonough2000) pointed out, learners receive and notice as much feedback as they need for their given stage of development. Therefore, the finding of a decrease in feedback over the course of the program might be interpreted as a sign of development during these informal conversation groups in this study abroad context. Additionally, the results from the tailored quiz revealed a 48% (10 out of 21 items) success rate on remembering the target word or phrase several months after the abroad experience. Previous work utilizing tailored quizzes following corrective feedback has found success rates ranging from 30% to 60% for lexical items and grammatical forms (McDonough & Sunitham, Reference McDonough and Sunitham2009; Williams, Reference Williams2001). The finding from the present study indicates a potential relationship between successful uptake of language following corrective feedback in this context. However, as the excerpts above show, there is no clear indication as to what either explicit or implicit feedback or the combination contributed to this finding.
RQ2: What differences exist between the corrective feedback provided by native speakers versus nonnative peer interlocutors during study abroad conversation groups?
Of the total number of feedback moves, the majority were provided by the NSs (total = 132; Time 1: M = 10.04, SD = 3.05; Time 2: M = 14.5, SD = 3.7; Time 3: M = 5.5, SD = 4.73), though NNS peer feedback did occur (total = 44; Time 1: M = 4.4, SD = 3.43; Time 2: M = 2.75, SD = 1.70; Time 3: M = 2.75, SD = 2.75). The difference found between NS and NNS feedback provision per conversation group was statistically significant (F = 22.38, df = 2, p = .001, η2 = .69; see Figure 4), with NSs consistently providing the majority of the feedback throughout the 6 weeks of conversation groups.
FIGURE 4. Change in Feedback Providers per Conversation Group Over Time.
These findings are also supported by interview data. As indicated above, the interviewed students stated that they asked specifically to be corrected by the NSs. However, their reaction to peer feedback was met with mixed reviews. Giselle described how one student who often explicitly corrected her peers was perceived as annoying and rude by the others. She stated that she preferred to be corrected by an NS rather than a language learning peer. This can be seen in Excerpt 5, in the dialogue between Ellen and Molly and the NS Diego. Here in line 8, Ellen, who wanted to talk about policies, used the false cognate policía, which means police. She was corrected by her fellow study abroad student Molly in line 11 (“políticas”); however, Ellen continued on with the same error despite this peer feedback. Ellen only modified her output to the correct form in line 14 after receiving feedback from the NS in line 13.
Excerpt 5. NS Versus NNS Feedback
Note. NS = native speaker; NNS = nonnative speaker.
The discourse exhibited in Excerpt 5 is in line with previously documented findings that learners do not always attune to feedback provided by peers due to a lack of confidence in the peers’ linguistic capabilities (Yoshida, Reference Yoshida2008). Learners also sometimes exhibit a preference for feedback delivered by NSs over feedback delivered by peers, as was found in Chu (Reference Chu2013). However, in other instances when learners noticed that an NNS peer was stuck or struggling to communicate a word or phrase, all of the interviewed students agreed that they would jump in to assist, as can be seen in Excerpt 6.
Excerpt 6. NS and NNS Feedback
Note. NS = native speaker; NNS = nonnative speaker.
After hearing this example in the stimulated-recall interview, Jessica stated that she was really open to this type of feedback and appreciated assistance from her peers. One student, Holly, stated that she did not perceive this type of feedback as a correction, per se, but more supportive and collaborative in nature. She stated during the interview:
When it's your same language and obviously as [University] students we have such similar experiences, you get halfway through your sentence and someone will know where you're going with it. Everyone kind of knows what you're going for so you kind of just shout things out until you find the best way to say it.
This feeling was echoed by the six interviewed students who stated that NNS peers did not correct each other but rather described negotiation for meaning or peer feedback as “supporting each other,” “debating over which verb to use,” “helping nervous people speak,” “similar to helping a classmate,” “only to say, ‘did you mean this?’” and “only to help with a word or phrase.”
Four students described a sequence of strategies they would employ when there was a misunderstanding or they were struggling with a word or phrase: They would first ask their NNS peers if they knew the word, then attempt to work together to explain the word or phrase in Spanish, and finally, if they were still stuck, they would directly ask the NSs for a translation because many of them knew English. These examples were characterized by overlapping speech as the NNSs participants jumped in to assist their peers in explaining the concept in Spanish. In the scaffolding excerpt (Excerpt 7), Maria was describing a Keurig coffee maker to the NSs. After several attempts to explain the type of coffee maker to the NSs, Elisa provided an example in English to help distinguish the concept and work toward understanding on the part of the NSs.
Excerpt 7. Scaffolding Excerpt
Note. NS = native speaker; NNS = nonnative speaker.
As can be seen in lines 46 to 50, the three peer NNSs worked together, scaffolding off one another's ideas. Although not all suggestions are correct (“copa” instead of “taza” in lines 49 and 50), by negotiating with one another, the NS was able to understand Maria's original intended meaning.
RQ3: What is the role of the L1 in facilitating opportunities to engage with corrective feedback in these conversation groups?
The results from the analysis of L1 English use revealed that students and NSs used English at times throughout the course of the 6-week program (total of 163 c-units of L1 use). Both NSs and NNSs produced slightly more English utterances per conversation group at the beginning of the program (n = 72 c-units) than at the end (n = 46 c-units). However, this change was not statistically significant (F = .652, df = 2, p = .54, see Figure 5).
FIGURE 5. Changes in L1 (English) Use per Conversation Group Over Time.
The results from the interviews and stimulated recalls revealed mixed perceptions concerning the use of English during these conversation groups. Three students stated that they did not notice any English use at all during the conversations groups. Ashley said they would avoid English so as to not make the NS students feel awkward. Overall, the use of English as a strategy (rather than simple code-switching) led to mixed effects with regard to the opportunities to engage in corrective feedback and negotiation for meaning.
Excerpt 8. English Use for Cultural References
Note. NS = native speaker; NNS = nonnative speaker.
Two students mentioned that they would only use an English word or phrase for American cultural references, idioms, or jokes, as was the case in Excerpt 8. In this example, Holly's use of the English word “surge” in line 6 did not lead to an instantaneous translation on the part of Elena (the NS); instead, Holly continued to explain “surge” in Spanish, receiving feedback and further scaffolding from her peers, until Elena understood her message in line 16. The use of English here (and the explanation that follows) initiates a sequence of facilitative peer scaffolding and corrective feedback episodes that enabled Holly to explain her message more clearly. Holly recalled after hearing an example in her stimulated-recall clip that she used the English word “surge” when talking about the car-sharing app Uber because:
That was an example of . . . I just didn't know. “Surge prices” to me is like not a word, it's not the same thing as a “surge,” . . . it's like a brand word, which is why I was like “surge [in English]” and then tried to define it.
In this case, Elena (the NS) did not simply translate “surge,” and instead the study abroad students engaged in a negotiation for meaning to explain this cultural reference. The breakdown in communication that the use of “surge” triggered allowed all of the learners in the group to engage in opportunities for corrective feedback.
Excerpt 9. English use to Request Vocabulary
However, the use of English did not always produce the same types of negotiations for meaning. In one example (also seen previously in Excerpt 3), a request for a vocabulary translation resulted in explicit feedback from the NS in Excerpt 9. When Audrey was asked why she used English words instead of trying to talk around the word in Spanish during her stimulated-recall interview, she recalled:
Even though in that situation I broke the language pledge . . . it was more about having a conversation than technically saying things in Spanish.
Although Excerpt 9 demonstrates how L1 use promoted conversation by allowing Aubrey to continue her utterance, it did not promote further negotiation for meaning. Maria described a similar feeling after listening to a clip of her asking for the word “expelled,” as shown in Excerpt 10.
Excerpt 10. English Use to Request Vocabulary
She added that she would use English so that she did not drag out a conversation too long looking for the right way to explain a concept:
If I had no idea what the word is or similar words to it, I would say it in English. I always feel bad when I'm taking too long speaking . . . trying to explain something that I could say in English really quickly. Trying to not monopolize any conversation. Like, just speed this up, someone will know this word.
Both excerpts indicate that learners utilized their L1 judiciously to continue the flow of meaning making in their conversation. The frequent use of the vocabulary request strategy allowed NSs to offer quick feedback so that the learners could make themselves better understood. However, it is clear that the vocabulary translation, while effective for keeping the conversation moving, did not initiate the same type of negotiation for meaning as when the NS did not understand or did not provide a translation. While it is unclear which type of L1 use promoted the most effective environment for uptake, these results indicate a tendency for learners to engage more following a communication breakdown that is not immediately resolved by the NS providing a translation or further instruction.
Overall, the learners in this study abroad program indicated that the little English that did occur in the conversation groups was used productively and encouraged the flow of conversation and meaning making, and therefore the learners were not in violation of the language pledge. As these data demonstrate, instances of L1 use that were not immediately resolved by an interlocutor enabled learners to engage in more opportunities to receive corrective feedback and negotiate for meaning with the NSs and their NNS peers.
LIMITATIONS AND FUTURE DIRECTIONS
There were several limitations to the current study that should be considered when reflecting on the generalizability of the results. First, the audio recordings obtained from the participants contained background noise, simultaneous conversations, and other sounds that were part of the natural environment that foregrounded the interactions but also, at times, obscured the data such that transcriptions could not be made. Similarly, although leaving recording in the hands of the participants themselves minimized intrusion and preserved the naturalistic setting, it also resulted in data loss, as not every participant was equally represented. It is also necessary to consider the possibility of the observer's paradox (Labov, Reference Labov1972), given that learners knew they were being recorded by their study abroad program, which could have affected how often they used their L1 or otherwise interacted.
While coding these recordings, only corrective feedback moves were transcribed, and therefore there are no total word counts for each conversation. Therefore, the amount of corrective feedback found in these data does not control for pauses or the varying recording times of the conversations that could have affected overall corrective feedback or English L1 use counts. Future iterations of this project should involve more tightly controlled recording requirements, possibly with the addition of video so that gesture and paralinguistic information could be obtained, as some previous research has indicated that gestures coupled with corrective feedback can impact development (Nakatsukasa, Reference Nakatsukasa2016). Video recordings would also enhance the clarity of some interactions and would make it easier to track particular participants. New innovations in technology such as unobtrusive 360 degree cameras, which sit in one place on a table yet capture everything happening around it, might be used in future studies.
The time lag of 4 months between the students’ sojourn and the qualitative and quantitative follow-up interviews and quizzes also significantly limited the validity of the post–study abroad data. While this was unavoidable due to geographic and time-related constraints, memory decay and potential exposure to the target language post–study abroad would likely have interfered with the perceptions that the learners shared during the interviews and their achievement on the tailored quizzes. Future studies should aim to collect stimulated-recall data as close to the interactions as possible, in keeping with the recommendations suggested by Gass and Mackey (Reference Gass and Mackey2015b).
Another limitation of the study concerns the personalities and relationships of the interlocutors. First, it is possible that some NSs or peers (such as the heritage speakers) were more willing to provide feedback than others, as previous research has shown that more confident learners are more willing to interact (Ranta & Meckelborg, Reference Ranta and Meckelborg2013). The NSs also had various levels of English proficiency. This would lead to unbalanced data in the various recordings. Sato and Lyster (Reference Sato and Lyster2012) trained their learners to offer peer-to-peer feedback; however, the learners in this study were not explicitly told to offer feedback. For these reasons, some recordings might contain learners who were more willing to provide feedback. Additionally, the amount of times individual learners provided feedback versus received feedback, either from the NS or the NNSs, was not controlled for in this study. Previous research has shown that learners benefit from providing feedback as well as receiving it (Sippel & Jackson, Reference Sippel and Jackson2015).
This research could profitably be replicated with study abroad students of different cultural backgrounds, as some cultures find providing peer feedback face-threatening (Sato & Lyster, Reference Sato, Lyster and Mackey2007; Fujii & Mackey, Reference Fujii and Mackey2009) or socially inappropriate. The current study also was located in a short-term study abroad program and involved interactions with NSs peers with proficiency in the L1 (English) of the study abroad students. This unique context makes the findings of this study difficult to generalize to other contexts. Replications with different interlocutors such as in service encounters or with interlocutors without proficiency in the students’ L1 are needed to enhance generalizability. Future studies might also compare these informal conversation groups with contexts of interaction such as classrooms abroad, homestays abroad, or laboratory task experiments to determine how the interactions compare in these diverse settings.
CONCLUSION
The implications for this study are twofold. First, in the field of SLA, corrective feedback is understudied in naturalistic settings. Therefore, this study offers a new contribution to the literature, which until now has focused almost exclusively on the types and targets of corrective feedback in classrooms or laboratory settings. Furthermore, peer feedback, an understudied type of feedback in and of itself, has until now not been studied in study abroad settings. This study offers a first look into learners’ engagement with corrective feedback and negotiation for meaning while abroad, a foundation that could be used to examine future theoretical and empirical questions.
Second, for practitioners in the study abroad field, this study offers a window into the conversations their students have with NSs during their stays in the host countries. Practitioners, and potentially students themselves, may have preconceptions about these informal conversation groups, namely, that feedback would not be present due to the focus on meaning making and pure socialization. However, the data clearly show that the rich input and modified output opportunities provided by these conversation groups go above and beyond the expectations of these simple meet-ups.
Finally, the use of study abroad students’ L1 while interacting in naturalistic settings, especially students who have taken a language pledge, remains underinvestigated. While the current study has uncovered the productive use of the L1 to engage in negotiation for meaning and corrective feedback episodes, future studies should examine which types of L1 use best promote opportunities for feedback and negotiation for meaning and make explicit connections to noticing and subsequent development. Practitioners of study abroad programs might consider developing workshops or other materials that show learners what kinds of L1 strategies best promote language development in their context. They might also show learners how to make the most of the corrective feedback they receive while interacting with NSs and peers throughout their sojourn. Overall, study abroad practitioners can have confidence that conversation groups are not just fun for study abroad students but also offer rich opportunities for learners to continue to develop their language skills.
Acknowledgments
The study is part of the BarSA Project developed by Dr. Cristina Sanz with Georgetown University and Language Learning funds. Thank you to the students of the Georgetown @ Barcelona summer program for their participation in data collection, as well as Natalia Curto Garcia-Nieto, Janire Zalbidea, and Brandon Tullock, who collected the data in Barcelona. A big thank you to the students in Dr. Sanz's seminar on study abroad and L2 development who commented on an earlier version of this work. Thank you to Dr. Alison Mackey, Dr. Luke Plonsky, and the three anonymous reviewers who provided feedback on this project.
