Studying abroad during and before the COVID-19 pandemic A comparison of target language use and self-reported linguistic progress

The study abroad (SA) context is assumed to be ideal for second language (L2) learning because of the opportunities it provides to use the target language (TL) and to socialize with TL speakers. Although research conducted before the COVID-19 pandemic generally confirms this assumption, little is known about how the COVID-19 pandemic modified conditions for L2 learning in a SA context. This study contributes by documenting Swedish French language students’ patterns of TL use and perceived linguistic progress in a SA context in France during the COVID-19 pandemic (spring 2021) and compares these data with data from an equivalent group of Swedish French language students who studied in France before the pandemic. The results suggest that the SA context provided advantageous learning conditions even during the pandemic.


Introduction
The study abroad (SA) context has a reputation of being ideal for second language (L2) learning as it provides learners the opportunity to interact with the target language (TL) -a precondition for L2 development -through participation in social and cultural life (e.g., Ellis, 2015;Pérez-Vidal, 2017).This feature presumably distinguishes the SA context from the foreign language learning context, where TL contact is assumed to be more or less restricted to the classroom.Indeed, several studies have shown that the SA context offers optimal opportunities for frequent TL use in a variety of contexts, although not all students take advantage of this opportunity (e.g., McManus et al., 2014).However, in the beginning of 2020, the world was struck by the coronavirus resulting in the COVID-19 pandemic (World Health Organization, 2020).This led nations to restrict social and cultural life.In certain countries, for example in France, public venues such as museums, gyms, movie theaters, restaurants, and bars were temporarily shut down to reduce the spread of the virus (e.g., French Government, n.d.).Since these spaces are common venues for social encounters and interaction, it seems possible that such a pandemic would affect language learning conditions for students studying abroad.Recently, some studies have been published that investigate the impact of the COVID-19 pandemic on various aspects of international education (e.g., Gaitanidis, 2020;Liu & Shirley, 2021;Ma & Miller, 2020).However, to the best of our knowledge, no study has investigated how COVID-19 restrictions have affected TL use and learning in the SA context.The purpose of this study is to contribute such knowledge.The study compares frequency and type of TL use outside the French language classroom and self-reported linguistic progress among Swedish students who studied in France during the COVID-19 pandemic with a similar cohort of Swedish students who studied in France before the pandemic.

TL use in SA
The SA learning context is assumed to be particularly favorable for L2 learning in comparison with the foreign language learning context, because learners have access to the TL outside the classroom (e.g., Segalowitz & Freed, 2004).To investigate the validity of this assumption, numerous pre-COVID-19 studies documented students' patterns of TL use, typically through questionnaires such as the Language Contact Profile (LCP) (Freed et al., 2004) or the Language Engagement Questionnaire (LEQ) (McManus et al., 2014), but also through interviews with sojourners (e.g., Mitchell et al., 2017).Several studies approached the question of TL use by SA students from both a quantitative (frequency or amount of TL use) and qualitative perspective (nature of TL use), and some combined these two, by presenting how frequently international students used the TL for various purposes during their stay in the TL country.In general, these studies showed that the SA context presents possibilities for frequent TL use in a variety of ways (passive and active) and often for interactive purposes (e.g., shorter and longer conversations, service encounters, text messaging), at least for those who took advantage of this opportunity (e.g., Hernández, 2010;McManus et al., 2014;Mitchell et al., 2017; for a recent overview, see Briggs Baffoe-Djan & Zhou, 2021).For example, Hernández (2010), using a modified version of the LCP to investigate how often 20 US students used Spanish during a semester in Spain, found that these students used Spanish outside the class between 31.5 and 115.5 hours a week (M = 60.68 hours, SD = 24.98).By far, the students used the TL most often to speak Spanish with first language (L1) Spanish speakers, although substantial individual variation was evident.Other frequent activities in Spanish included writing homework assignments, listening to music, and listening to TV and radio.Another example of a research project that thoroughly mapped out international students' patterns of TL both through the survey and interview method, is the LANGSNAP project (see Mitchell et al., 2017).McManus et al. (2014), scrutinizing one part of LANGSNAP, investigated how 29 British students engaged in activities in French during a year abroad in France.The participants resided in 16 different regions in France and were either enrolled in classes at a French university (n = 8) or engaged in a work placement (n = 6) or as a teaching assistant (n = 15).The reported data were collected at three times during the year abroad.The five most frequent activities reported in French were "small talk, " "service encounters, " "long casual conversations, " "browse internet, " and "read text messages." In other words, it appears that the British students mainly used the TL for interactive purposes.
In a related study of 41 Swedish university students spending a semester in France, Arvidsson (2019) used the LEQ to gather information on the participants' language practices to examine whether the frequency of TL use predicted gains in language proficiency, focusing on multiword expressions.The LEQ was used to gather information on the participants' language practices and an overall mean frequency score was calculated for TL use by summing frequency scores for each LEQ item and then dividing the sum by the number of language-related activities.The frequency score ranged from 0 (never) to 5 (every day).The mean figure for all of the activities included in the LEQ was 2.07 (SD = 0.55).A closer scrutiny of the data revealed that the Swedes most frequently used French in service encounters, to read literature, and to browse the internet at least several times a week on average.The study included sub-groups of students who were enrolled in a French language program through three different Swedish universities.One of the subgroups (n = 25) directly matches the sample of the present study.As described in Section 3, the data coming from that sub-group were used in the present study, to compare patterns of language use among Swedish students who studied in France during the COVID-19 pandemic and Swedish students who studied in France before the COVID-19 pandemic.
The above-mentioned studies suggest that the SA context provides rich opportunities for social interaction in the SA, be it through service encounters, small talk with strangers/acquaintances in bars, and longer sustained conversations with new friends.According to Segalowitz and Freed (2004), the access to social interaction through TL speakers 1 distinguishes the SA context from the foreign language learning context.Social interaction is crucial for L2 development, as it provides the learner with meaningful TL input, scaffolding, and feedback (for an overview, see Ellis, 2015).Social interaction in the TL may also promote intercultural learning, including an increased understanding for how language is used in social situations (for an overview of the "language and intercultural dimension of education abroad, " see Jackson, 2020).Given that social interaction is important for L2 learning, and that social networks in SA partly determine the language learner's access to interaction in the TL, recent research has investigated international students' social networks (e.g., Dewey et al., 2012Dewey et al., , 2013;;Isabelli-García, 2006;Mitchell et al., 2017;Paradowski et al., 2021; for a recent overview, see Briggs Baffoe-Djan & Zhou, 2021).Such studies also used questionnaires and interviews to investigate the social networks of SA students, and some studies examined how social networks with TL speakers are formed.Typically, these studies found that some international students make acquaintances and even friends with TL speakers although some tend to make friends exclusively with co-nationals or other international students.For example, fewer than half (12 of 29) of the British SA students participating in LANGSNAP made friends with TL speakers during their year abroad (McManus et al., 2014).In fact, interviews revealed that SA students often experience difficulties establishing social relations with TL speakers, a difficulty that they did not necessarily anticipate (e.g., Arvidsson, 2021).Some scholars argue that individual factors such as personality or foreign language anxiety, as well as structural factors, influence the extent that a student seeks opportunities to use the TL outside of class and creates social ties with locals or other TL speakers (e.g., DeKeyser, 2014;Paradowski et al., 2021;Pérez-Vidal, 2014).
The social network research has also revealed that social and cultural life in public spaces are important platforms for connecting with locals and other TL speakers.Scholars have asked international students how they came in contact with locals or other TL speakers.Mitchell et al. (2017) found that the British students in France mainly made French-speaking acquaintances at university.Travelling, going out, and organized free time, including sport activities, also fostered social networking in the TL.Dewey et al. (2013) asked 71 US learners of Arabic what they had done to come in contact with TL speakers during their SA in Egypt or Jordan.The most frequent strategies included "striking up conversation with unknown individuals in public places" and "meeting new friends through existing friends" (p.269), suggesting the importance of public space and social gatherings for the kind of contact formation that leads to social interaction in the TL.This was also found in Arvidsson's (2021) case study based on interviews with Swedish SA students.For example, Elsa (the names of participants have been changed to protect their anonymity) joined a sports team at the beginning of her SA in France.Practice sessions and other social activities with her team allowed her to establish a solid social network that included several TL speakers.Other students in that same Swedish sample struck up conversations with French-speaking people in bars (Gustaf ) or organized dates through the dating application Tinder (Nova) to create occasions to speak French.For pre-pandemic language students, there were plenty of opportunities to socialize in the TL in the SA context, although it appears that it is up to the individual learner to create these opportunities.

Language learning in SA
Evidence gathered pre-COVID-19 suggests that the SA context is beneficial for language learning, at least for students who take advantage of the opportunities (e.g., Dewaele & Dewaele, 2021;Dewey et al., 2012;Howard, 2005;Mitchell et al., 2017;Regan et al., 2009;Segalowitz & Freed, 2004).Using objective (e.g., Dewaele & Dewaele, 2021;Howard, 2005;Mitchell et al., 2017;Regan et al., 2009;Segalowitz & Freed, 2004) and/or subjective (e.g., Dewaele & Dewaele, 2021;Dewey et al., 2012) assessment methods, various studies have documented the linguistic development of SA students.In general, the SA context appears to be particularly beneficial for the development of L2 competencies related to social interaction, including oral proficiency and pragmatic and sociolinguistic competence (e.g., Kinginger, 2013).These benefits are explained by reference to the characteristic feature of the SA learning context -i.e., students' exposure to TL use in informal social situations.Some studies compared language learning outcomes between a SA and an at home (AH) cohort.Although not all studies point in the same direction, the evidence suggests that L2 learning in the SA context is more effective than the AH context for the development of certain L2 skills.For example, Segalowitz and Freed (2004) observed that the 22 American students who studied L2 Spanish in an SA context made greater gains in oral fluency than their 18 peers who studied L2 Spanish in an AH context.Similarly, Regan et al. (2009) found that the SA context led to greater gains than the AH context in sociolinguistic competence for 72 Irish L2 French learners, and Howard (2005) found that the SA context was more conducive than the AH context to sociolinguistic development for Irish (n = 12) learners of L2 French.
Whereas these studies used objective assessment methods, other studies employed subjective assessment methods to assess proficiency gains.Students generally feel that they develop their L2 skills while studying abroad.For example, Dewey et al. (2012) asked 204 American SA learners to self-report their development relating to a number of speaking tasks in L2 Japanese after their SA in Japan (mean length of SA = 8.4 months).On average, the students had studied Japanese for 2.07 (SD = 1.87) academic years before arriving in Japan.During their stay in Japan, they perceived having developed their capacity to use the L2 mainly for personal needs as well as their ability to describe and to narrate in the L2.
Dewaele and Dewaele (2021) investigated both actual and self-reported proficiency among 33 British students who spent 4 to 12 months in a foreign country.Actual and self-reported proficiency were assessed at three times: before, during, and after the SA.At each time, the participants took a vocabulary size test and were asked to self-report their proficiency in writing, reading, listening, speaking, pronunciation, vocabulary, and grammar on a scale from 0 to 10.The authors found that the participants significantly increased their objectively measured proficiency in SA as well as their self-reported proficiency in all these domains, except for pronunciation (Dewaele & Dewaele, 2021).The authors also found significant relationships between actual and self-reported proficiency measures, indicating that the international students' perceptions of L2 development in SA tended to coincide with objective evaluations.Arvidsson et al. (2018) also used the self-report method to assess linguistic progress in the speaking abilities of 59 Swedish and Belgian intermediate students who studied in different countries.They explored relationships between TL use and the development of multicultural effectiveness and between self-perceived speaking progress and the development of multicultural effectiveness.With respect to linguistic progress, the students were asked to indicate if they had the impression of having made no progress, moderate progress, or considerable progress in speaking during their stay abroad.On average, the 59 Swedish and Belgian students felt they made moderate to considerable progress, as indicated by the mean score of 1.53 on a scale ranging from 0 to 2 (SD = 0.57).
Although the above-mentioned studies show that the SA context may confer rather ideal conditions for L2 learning, they were conducted before the COVID-19 pandemic.From its onset in March 2020 (World Health Organization, 2020), the pandemic dramatically impacted international education, including the cancellation of many SA programs (e.g., NAFSA, 2020).Some institutions gradually restarted their SA programs despite the continuation of the pandemic.To the best of our knowledge, no study has investigated the SA experience during COVID-19 from a language learning perspective, a gap that this study hopes to fill.Before presenting the study and the findings, I will briefly present the French SA context during the pandemic as this constituted rather different conditions than the prepandemic conditions.

The French SA context during a pandemic
Since the pandemic started, France took measures to decrease the spread of the COVID-19 virus (French Government, n.d.).Some of these measures were still in place during the spring semester 2021, which largely impacted social and cultural life, as well as citizens' right to move freely outside their homes.For example, between January and May 19, restaurants, bars, and cafés were closed as well as cultural sites, including libraries, museums, concert venues, theaters, and movie theaters (French Government, n.d.).In addition, a curfew between 6:00 p.m. 2 and 6:00 a.m. was in place; that is, people were not allowed to be outside their homes during the curfew (with the exception for specific instances, which required filling out a form supplied by the government) (French Government, n.d.).On March 31, the French president announced a four-week full stay at home-order; everyone was required to stay at home with the exception of certain activities such as food shopping and going for walks within a one kilometer of home (French Government, n.d.).After May 19, restrictions loosened up.Restaurants, bars, and cafés with outdoor seating re-opened and the curfew was amended to start at 9:00 p.m. rather than 6:00 p.m. (French Government, n.d.).During this semester, everyone was required to wear a face mask in public spaces (French Government, n.d.), including in class.It is, to date, unknown to what extent the pandemicrelated restrictions affected the L2 learning conditions for SA students, which brings us to the present study.

The study
This exploratory study investigates conditions for TL use and learning in the SA context during the COVID-19 pandemic.To this end, this study draws on data collected from Swedish French language students who studied in France during the COVID-19 pandemic (spring 2021) and compares these data with data collected from a similar cohort before the COVID-19 pandemic (fall 2016 and spring 2017).The comparison data were partly used for other research purposes (Arvidsson, 2019), with data from other sub-cohorts.The study addresses the following research questions (RQ):

Regional variations
Studying abroad during and before the COVID-19 pandemic RQ1.Compared to pre-pandemic SA students, how frequently and in what ways did the Swedish French language students use the TL while in France?RQ2.Compared to pre-pandemic SA students, to what extent did the Swedish French language students perceive themselves making TL progress while in France?

Participants
The participants in this study are Swedish French language students who studied in France for one semester, either during the pandemic (n = 10) or before the pandemic (n = 25).The two groups are referred to as SA-in-pandemic and SA-prepandemic, respectively.Both groups followed the same French language program through a Swedish university, with an entry level requirement corresponding to the A2 level according to the Common European Framework of Reference for Languages (Council of Europe, 2001).That is, the students were studying intermediate-level L2 French.The program includes modules for literature, grammar, oral proficiency, and French culture and society.During the semester, both groups studied fulltime (100%) and received about 10 hours of guided learning per week.That is, both groups were thus given classes on campus in a French city, with the exception of eight lessons which were given online in the SA-inpandemic sample, due to circumstances related to COVID-19.
The mean age was similar in both groups: the mean age of the SA-inpandemic group was 21.10 (SD = 2.99) and ranged between 19 and 29 and the mean age of the SA-pre-pandemic group was 21.00 (SD = 2.73) and ranged between 19 and 26.The French language was an L2 for all the students (one of the SA-in-pandemic participants was excluded as she reported having French as her L1).

Data overview and data collection procedure
This study draws on survey data which were collected by the researcher during spring 2021 to investigate learning conditions in the SA context during the pandemic and during fall 2016 and spring 2017 to investigate other research questions (see Arvidsson, 2019, described above).In both samples, a survey was distributed to the students at the end of their semester abroad.The SA-in-pandemic students were sent a survey through the web-based software Survey and Report.The SApre-pandemic students were given a paper survey (see Arvidsson, 2019).Identical datasets from SA-in-pandemic and the SA-pre-pandemic were prepared, enabling a comparison of the two samples according to demographics, patterns of TL use, and self-reported linguistic progress.These data are described in Section 3.3.

Data
The survey mentioned above yielded information on demographics (age, sex, and language learning experience), TL use, and self-reported linguistic progress.3TL use was assessed through a slightly modified version of the LEQ (McManus et al., 2014).This questionnaire asked respondents to estimate how often they had engaged in TL activities (i.e., French): never (0), rarely (1), a couple of times a month (2), a few times a week (3), several times a week (4), or every day (5).The original version contained 27 activities, but the present study draws on just the 20 items that asked about TL use outside the classroom, the focus of this study.
Regarding self-reported linguistic progress, respondents were asked to report their perceived linguistic progress with respect to speaking, writing, listening, and reading in L2 French: (1) "I have the impression that I have not made any progress during my stay"; (2) "I have the impression that I have made moderate progress during my stay"; and (3) "I have the impression that I have made considerable progress during my stay." Although such a method does not assess actual proficiency development, previous studies have found significant agreement between self-reported and actual proficiency, suggesting that L2 users relatively accurately self-evaluate their language proficiency (Dewaele & Dewaele, 2021;Liu & Brantmeier, 2019).

Data analysis
The data were imported to the software SPSS version 27.Descriptive and inferential statistics were used to respond to the research questions.To examine frequency of TL use (RQ1), I referred to group mean scores for total TL use and group mean scores for each of the 20 individual TL activities listed in the LEQ.The mean score for total TL use was obtained by summing the participant's responses to the 20 items (see Arvidsson, 2019;Mitchell et al., 2017) and dividing this sum by 20.The mean score ranges from 0 to 5 (0 = never and 5 = every day).To examine how SA-in-pandemic students used the TL French throughout the semester compared to SA-pre-pandemic students (RQ1), I compared the two samples' mean frequency scores for the most frequent LEQ activities (i.e., LEQ items with a mean score greater than 3, which corresponds to a few times a week or more often), the activities involving face-to-face social interaction, since the SA context is assumed to be particularly beneficial in this regard (e.g., Segalowitz & Freed, 2004), and the LEQ activities where the group mean differences differ significantly (see information on the statistical testing below).
A visual inspection of the data distribution reveals that the data were far from normally distributed.As bootstrap samples are less sensitive to the distribution of the data, the bootstrap method was used (see Larson-Hall & Herrington, 2010).To compare the SA-in-pandemic and SA-pre-pandemic mean scores for TL use, independent t-tests were conducted on a stratified bootstrapped sample using 1000 randomly generated replications with Group designated as the strata variable.
With respect to self-reported linguistic progress, a descriptive analysis indicated that students only used the two upper response options (2 and 3) of the three available options.Therefore, this variable was treated as categorical ("I have the impression of having made moderate progress" vs. "I have the impression of having made considerable progress").A contingency table was created to test the association between the two categorical variables: Group (SA-in-pandemic and SA-pre-pandemic) and "Self-reported linguistic progress" (moderate and considerable).Given that the data did not meet the assumption of cell value (>5), I conducted Fisher's exact tests to test the association of the variables for each of the four skills (reading, writing, listening, and speaking).This analysis was also conducted on a stratified bootstrapped sample using 1000 randomly generated replications with Group designated as the strata variable.
Effect sizes are reported and interpreted according to Plonsky and Oswald's (2014) field-specific recommendations: small (d = .40),medium (d = .70),and large (d = 1.00).For all analyses, a 5% significance level was considered and, thus, results were significant if the corresponding p-values were less than 0.05.

Frequency and type of TL use
The first research question asked how frequently and in what way Swedish SA students used French during the COVID-19 pandemic.The mean figure for TL use in Table 1 indicates that the pandemic cohort engaged in French language activities with an average frequency of 2.19 on a scale from 0 to 5 (SD = 0.52), compared to an average frequency of 1.94 for the SA-pre-pandemic cohort (M = 1.27,SD = 0.74).However, an independent samples t-test based on a stratified bootstrap sample reveals that the mean difference between the groups is not significant (95% CI [−.12, .58],p = .214).According to Plonsky and Oswald's (2014) guidelines, the effect size for this comparison was small to medium (d = .53).For the LEQ items in Table 2 with a mean score of >3 (i.e., a few times a week or more often), the SA-in-pandemic group used French at least several times a week for six purposes: to engage in service encounters (M = 4.10, SD = 0.88, 95% CI [3.60, 4.50]); to browse online (M = 3.60, SD = 1.43, 95% CI [2.70,4.40]); to listen to music (M = 3.50, SD = 0.97, 95% CI [3.00, 4.10]); to engage in small talk (M = 3.50, SD = 1.72, 95% CI [2.40,4.40]); to read literature (M = 3.20, SD = 0.92, 95% CI [2.50, 3.70]); and for online social network platforms (M = 3.00, SD = 1.33 95% CI [2.20,3.80]).Only four activities were performed as frequently in the SA-pre-pandemic group, which were mentioned as often in the SA-in-pandemic group: service encounters (M = 4.24, SD = 0.66, 95% CI [4.00, 4.52]); to browse online (M = 3.29, SD = 1.27, 95% CI [2.83, 3.75]); to listen to music (M = 3.08, SD = 1.35, 95% CI [2.64, 3.60]); and to read literature (M = 3.24, SD = 0.88, 95% CI [2.88, 3.56]).As indicated by the test statistics in Table 3, the mean scores related to these frequent activities did not differ significantly between the groups, suggesting that the SA-in-pandemic and the SA-pre-pandemic groups used the TL equally as often and in almost identical ways during their studies in France.
Focusing on the TL activities that involve face-to-face social interaction, Table 3 reveals that there were no significant mean score differences between the groups with respect to service encounters and small talk but that the SA-inpandemic group used TL French significantly more often than the pre-pandemic group to engage in informal conversations, resulting in a large effect size (d = 1.41) according to Plonsky and Oswald's (2014) benchmark for second language acquisition.The mean scores in Table 2 reveal that the SA-in-pandemic group used the TL in informal conversation almost a few times a week (M = 2.90, SD = 1.73, 95% CI [1.90.4.00]), whereas the SA-pre-pandemic group rarely engaged in informal conversations in TL French (M = 1.16,SD = 0.99, 95% CI [.84, 1.60]).
Finally, an inspection of the test statistics for mean differences related to all 20 LEQ items presented in Table 3, reveals that the SA-in-pandemic cohort used the TL significantly more often than the SA-pre-pandemic cohort, not only in longer informal conversations, but also to read and write text messages, although significantly less often to read emails, journals, and magazines.However, the mean frequency scores for these activities are generally low in both groups (see Table 2), meaning that neither the SA-in-pandemic cohort nor the SA-pre-pandemic cohort used the TL much for these purposes.It should also be noted that, in general, there is certain individual variation in mean scores, which is indicated by the SD figures in Table 2.

Self-reported linguistic progress
Research question 2 concerns the extent that both cohorts perceived making linguistic progress during their semester abroad.Table 4 lists the distribution of self-reported linguistic progress for reading, writing, listening, and speaking.The response options included "no progress, " "moderate progress, " and "considerable progress." However, all students, regardless of cohort, felt they had made progress to some extent as they all reported having made "moderate" or "considerable" progress in all four skills during their semester abroad.Table 4 also reveals that the SA-in-pandemic students responded in relatively similar ways as the SA-prepandemic students.That is, most students in both cohorts felt they made "considerable" rather than "moderate" progress in all four skills throughout the semester, although this pattern was less clear for speaking than for the other skills.In the SA-in-pandemic cohort, four students estimated they had made "moderate" progress and six estimated they had made "considerable" progress during their semester in France.This pattern is repeated in the SA-pre-pandemic group: eight students estimated they had made "moderate" speaking progress and 17 estimated they had made "considerable" speaking progress.A series of Fisher's exact tests reveal that there is no significant association between the progress in any variable and cohort for any of the four skills: reading (phi = .28,p = 1.00), writing (phi = .16,p = .447),listening (phi = .12,p = .64),or speaking (phi = −.08,p = .706).
In other words, the SA-in-pandemic students perceived making as much linguistic progress as the SA-pre-pandemic students.Studying abroad during and before the COVID-19 pandemic

Discussion
This study investigated conditions for TL use and learning in the SA context during the COVID-19 pandemic by comparing a cohort of Swedish French language students studying in France during the COVID-19 pandemic (SA-in-pandemic) with a matched cohort taking the same French language program in France before the COVID-19 pandemic (SA-pre-pandemic).For the SA-in-pandemic cohort, data were collected near the end of spring 2021.These data were compared with SA-pre-pandemic data collected by the researcher for other research questions.
To address the first research question, this study compared how frequently and for what purposes French language students from Sweden used the TL while in France during the pandemic with how frequently and for what purposes French language students from Sweden used the TL while in France before the pandemic.The data consisted of responses to a modified version of the LEQ (McManus et al., 2014), where students estimate how frequently they used the TL for different purposes on a scale from 0 (never) to 5 (every day).No significant differences were found between the SA-in-pandemic cohort and the SA-prepandemic cohort.That is, SA-in-pandemic students used the TL as frequently as the SA-pre-pandemic students.Although the overall mean frequency score is a rather crude indicator of frequency of TL use, this finding suggests that the sheer frequency of TL use was not impacted by the restrictions imposed on society during the pandemic.This finding is corroborated by further analysis of the data.To investigate how the SA students typically used French during their stay in France, I examined what activities the students engaged in at least a few times a week (i.e., >3, on the scale from 0-5): both cohorts used the TL in almost identical ways and the frequency figures did not differ significantly between the two groups.Specifically, on average, both cohorts used the TL several times a week to engage in service encounters (i.e., verbal exchanges in service-related situations) and a few times a week to browse online, listen to music, and read literature.In addition, the SA-in-pandemic cohort used the TL in small talk (i.e., shorter conversations about everyday matters) and in online social network platforms at least a few times a week, on average.Although the SA-pre-pandemic cohort used the TL less often for these purposes, there were no statistically significant differences.Overall, these observations suggest that the students engaged in TL activities relatively frequently and for various purposes even during the pandemic -an observation that heretofore had only been observed in pre-pandemic SA students (Arvidsson, 2019(Arvidsson, , 2021;;Hernández, 2010;McManus et al., 2014;Mitchell et al., 2017).
It is surprising that the patterns of TL use did not differ in any important ways between the SA-pre-pandemic and SA-in-pandemic samples.As described in Section 2.3, there were restrictions in place in French society when the SA-in-pandemic students were in France, including an order to stay at home during given hours of the day.One could assume that the time spent in the home environment would lead the SA-in-pandemic sample to engage more in private TL activities (i.e., reading, listening) than their SA-pre-pandemic peers, yet such an assumption does not find support in the data.Both cohorts spent a fair amount of time reading and listening in the TL throughout their stay abroad regardless of whether their SA took place during the pandemic or not.
Second, one could assume that the pandemic-related restrictions in place during spring 2021 would limit, rather than promote, the SA-in-pandemic students' opportunities to engage in both service encounters and in small talk in the TL.Despite this, the in-pandemic sample reported finding plenty of opportunities to use the TL for such interactions.Where did such interactions take place?Before the pandemic, such spontaneous everyday interactions seemed to take place as international students visited stores, cafes, bars, restaurants, and gyms (e.g., Arvidsson, 2021).Most of such kind of venues were closed during spring 2021 except for supermarkets, which apparently constituted an important platform for human contact during the pandemic (Franceinfo, 2021).It is likely that this is where the SA-in-pandemic students' frequent service encounters took place during spring 2021, and possibly where they ended up engaging in small talk with others.Some supermarkets only received limited amounts of customers at a time, which in some cases created lines outside the supermarket (Franceinfo, 2020) where small talk between costumers may also have taken place, although the data do not allow us to draw any such conclusions.
The in-pandemic sample's frequent use of online social network platforms is however less surprising.For those who had internet access in their homes, online activities were not affected by the COVID-19 restrictions, but -perhaps -rather promoted by them.As we shall see, this type of TL engagement may help explain the finding related to the students' engagement in longer informal conversations.
A closer scrutiny of the students' face-to-face social interactions revealed that, not only did the SA-in-pandemic group frequently participate in small talk and service encounters, but they used French significantly more often in longer informal conversations than the SA-pre-pandemic cohort.While the SA-in-pandemic students sustained longer informal conversations in French almost a few times a week on average, the SA-pre-pandemic rarely engaged in longer conversations in the TL (between rarely and a couple of times a month).The effect size for this difference was large.This finding certainly aligns with research conducted before the pandemic, which has shown that the SA context offers rich opportunities for social interaction in the TL (e.g., Hernández, 2010;Mitchell et al., 2017;Segalowitz & Freed, 2004).However, it is both interesting and surprising that such opportunities appeared to have been stable even during the pandemic.In other words, the conditions for social interaction in the TL do not seem to have been negatively impacted by the pandemic, despite certain public spaces being shut down and despite mandates to wear face masks in public spaces (French Government, n.d.).The data set does not tell us how opportunities to engage in longer conversations were created, nor in what situations such conversations took place.Pre-pandemic research has found that private social gatherings and organized social activities in public venues (e.g., sport or music) constitute platforms for students to make contact with locals and make friends, and in this way to sustain longer conversations with TL speakers (e.g., Arvidsson, 2021;Dewey et al., 2013;Mitchell et al., 2017).The COVID-19 restrictions mentioned above, however, implied reduced possibilities to organize social gatherings and activities.It is possible that the SA-in-pandemic students' above-mentioned use of online social network platforms helped them to come in contact with TL speakers -a strategy used by other Swedish French language students in France before the pandemic (see Arvidsson, 2021).Perhaps such strategies were even more frequent during the pandemic than before, although these remain speculations.
Despite the COVID-19 restrictions, the students who went abroad during the pandemic thus clearly managed to find opportunities to interact in French.This in itself suggests that these students were generally adaptive and resourceful.Scholars studying pre-pandemic SA contexts have put forth that individual factors such as motivation, personality, agency, and foreign language anxiety play a role in creating opportunities for social interaction (e.g., DeKeyser, 2014;Hernández, 2010;Paradowski et al., 2021;Pérez-Vidal, 2014).It is possible that such factors played an even more pronounced role during the pandemic, when societal restrictions reduced contact opportunities.However, further research is needed to test these assumptions.
The analyses revealed that, apart from the significant difference related to longer informal conversations, mean scores differed significantly only for four other items.The SA-in-pandemic cohort used the TL more often than the SApre-pandemic cohort to read and write text messages.This may indicate that the SA-in-pandemic students sustained contact with TL speakers to a different extent than the pre-pandemic sample -which in turn would help explain that they had more informal conversations than the SA-pre-pandemic sample, as suggested above.The SA-in-pandemic cohort, however, used the TL less often than the SA-pre-pandemic cohort to read emails, journals, and magazines, although mean frequency scores for these activities were relatively low in both cohorts.Taken together, these data also indicate that the COVID-19 restrictions may not have hindered the SA-in-pandemic students from using the TL in similar ways as the SA-pre-pandemic students.Only through qualitative data, however, could we better understand the ways in which the COVID-19 restrictions affected students' patterns of TL use and their social relationships.
Finally, there was certain individual variation related to the LEQ items in both cohorts.This finding is not surprising and aligns with studies conducted before the pandemic (e.g., Arvidsson, 2021;Hernández, 2010;Mitchell et al., 2017).Individual differences of the kind mentioned above, such as motivation and personality-related factors, may help explain why different students engage in different activities to varying extents.
Research question 2 addressed the students' perceived linguistic progress related to reading, writing, speaking, and listening.Students were asked to selfreport their perceptions of progress on a scale from 1 to 3 ("no progress, " "moderate progress, " and "considerable progress").The data, which were treated categorically, revealed that all students in both cohorts believed that they made progress in all four skills.In both cohorts, most students felt they made "considerable" rather than "moderate" progress and there were no statistically significant differences in response patterns -that is, the self-assessed progress of the inpandemic students and the pre-pandemic students was similar.The observation that the SA context is conducive to L2 development is in line with research conducted pre-COVID-19 using objective or subjective assessment methods, including participants with varying pre-SA proficiency levels (e.g., Arvidsson et al., 2018;Dewaele & Dewaele, 2021;Dewey et al., 2012;Howard, 2005;Regan et al., 2009;Segalowitz & Freed, 2004).This study, however, adds to previous research by suggesting that the SA context was beneficial for L2 speaking development even during pandemic restrictions.
Mirroring the discussion regarding the students' patterns of TL use, it is somewhat surprising that SA in-pandemic did not promote the students' reading and listening proficiency in French more so than their speaking proficiency.In theory, it could again be assumed that the COVID-19 restrictions on social life would lead the in-pandemic students to practice and develop less social TL skills to a larger extent than their pre-pandemic peers.This assumption was, however, not born out in the data, which may be understood in relation to the observed similarities in the two cohorts' patterns of TL use.
Previous research has shown that the SA context is particularly beneficial for L2 speaking development (see Kinginger, 2013) because of the opportunities it offers to interact with TL speakers.It was assumed that the COVID-19 restrictions on social and cultural life could reduce opportunities to practice and consequently impede the SA-in-pandemic students' L2 speaking development.In actuality, they felt that they developed their L2 speaking skills as much as the SApre-pandemic students.While this is not surprising in light of the observation that the in-pandemic-students engaged more frequently in informal conversations than the pre-pandemic-students, it is surprising in that the pandemic did not seem to impact negatively students' access to social interaction and perceived speaking development.Future studies could use qualitative methods such as interviews or the diary method to gain insights into students' emotional and behavioral reactions to unexpected or less-than-ideal circumstances for TL use in the SA context, and to explore in what ways they create learning opportunities under such circumstances.
This study has its limitations.The questionnaire data only provide limited insights into the students' TL use and do not reveal any information regarding the context for their different TL activities, nor regarding the nature of their social interactions.In order to reach a more comprehensive understanding, the results would have to be complemented with qualitative data.In addition, the study is based on self-assessments of language learning progress, and the data do not allow for any conclusions regarding actual skill development in reading, writing, listening, and speaking.However, the validity of the findings is strengthened by the fact that the groups were comparable in terms of pre-departure proficiency and contributes information about the possibility to develop L2 skills during the COVID-19 pandemic.Furthermore, as the study concerns only one student population (Swedish university students studying French) and one SA context (France), it is impossible to generalize the findings to other SA populations and contexts.In addition, although robust methods were used in the statistical treatment of the data, the initial sample sizes were small, so the results should be interpreted with caution.Ideally, the study will be replicated in the same context and in a similar population after COVID-19 restrictions are lifted, which would allow for confirmation or rejection of the findings of this study.
In conclusion, the findings suggest that the SA context provided rich opportunities for TL use and L2 development even during pandemic restrictions.The study adds to recent research regarding the effects of the COVID-19 pandemic on international education (see e.g., Gaitanidis, 2020;Liu & Shirley, 2021;Ma & Miller, 2020) and specifically contributes knowledge about the conditions for L2 learning in the SA context during the COVID-19 pandemic.

Table 1 .
Mean frequency of TL use

Table 2 .
Mean frequency scores for LEQ items (Items with a mean frequency score of >3 are marked in bold)

Table 3 .
Test statistics *** for each of the LEQ items (results marked in bold are significant at the 95% confidence level)

Table 3 .
(continued) *** Independent sample t-tests performed on a bootstrap sample based on 1000 randomly generated replications with "Group" as strata variable.

Table 4 .
Contingency tables * for self-reported linguistic progress in SA-in-pandemic versus SA-pre-pandemic groups