Examining the role of learning context and individual differences in gains in L2 writing performance: the case of teenagers on an intensive study-abroad programme

ABSTRACT The present study examines the effects that a 3-week study-abroad (SA) experience and a set of individual differences have on the foreign language (FL) development of a group of 64 Catalan/Spanish speakers, learners of English as an FL. Moreover, the present study attempts to account for the outcome differences between successful and less successful FL learners. Results suggest that short SA experiences are beneficial for the FL development, that certain individual differences such as motivation and FL use in free time (among others) play a role, and that the successful and less successful learners’ SA experience differed in several aspects that could explain the differences in their outcome.


Introduction
The effects of the study-abroad (SA) learning context on second language (L2) acquisition have been an important focus of recent research (Llanes 2011), probably due to increasing globalisation of education and the associated importance of learning L2 English. Although there are many ways of learning an L2, going abroad to a country where the L2 is the official language is believed to be one of the most efficient (Collentine 2009;Freed 1995). This may account for the growth of SA programmes, especially short SA, in which thousands of students of different ages engage every year with the aim of improving their L2 proficiency.
The majority of the studies examining the effects of the SA learning context on L2 acquisition have focused on adult participants, in particular, university students spending a semester abroad (Freed 1995;Pérez-Vidal and Juan-Garau 2009). Very few studies document the L2 gains of adolescents studying abroad and, at the same time, very few studies examine the impact of short SA experiences. The lack of research on short SA experiences and on adolescents is surprising given the large numbers of adolescents engaging in short SA programmes every year. Another neglected area is the role of individual differences in accounting for the impact of SA experiences. While research has investigated L2 gains (or lack thereof) as a result of SA, the possible effects of individual differences such as participants' different types of motivation or attitudes towards the development of the L2 have not been extensively investigated. The present study, therefore, seeks to address these gaps by examining the impact on L2 writing scores of the SA learning context and a set of individual variables in a group of adolescent learners aged 12-17 after three weeks of English SA.

SA and L2 development
The effects of the SA context have been explored with respect to several L2 areas, with oral skills receiving the most attention. There are numerous studies which confirm that spending some time abroad has a positive impact on L2 oral skills (Lennon 1990; Llanes andMuñoz 2009, 2013;Martinsen 2010), and several studies (Hernández 2010;Llanes and Muñoz 2013) confirm the superiority in terms of oral skills development of the SA context over the 'at home' (AH) context (i.e. where participants study the L2 in their home country). Studies examining L2 vocabulary acquisition following an SA experience also seem to confirm the benefits of the SA context since participants typically increase their L2 lexicon significantly after spending some time abroad (Dewey 2008;Foster 2009;Milton and Meara 1995). Studies analysing the effects of SA on areas such as pragmatics and sociolinguistic skills also provide evidence in favour of the SA context (Regan 1995;Regan, Howard and Lemée 2009).
However, the clear benefits of SA identified for oral skills, vocabulary and pragmatics contrast with rather more mixed results for grammatical development. The studies that have examined L2 grammar development after SA have found that this area does not improve significantly whereas typically, participants studying in their AH country do show improvement (Collentine 2004;O'Donnell 2004). The effects of the SA context on other areas such as writing and pronunciation are also less clear-cut, with research presenting contradictory findings; for example, studies by Díaz-Campos (2004) and Mora (2008) do not find any significant improvement in the L2 pronunciation of their participants after spending some time abroad, whereas other studies such as Stevens (2001Stevens ( , 2011 and Muñoz and Llanes (2014) do.
In relation to L2 writing skills, some studies such as Sasaki (2009), Pérez-Vidal and Juan-Garau (2009) and Serrano, Tragant and Llanes (2012) show that writing skills do improve as a result of SA, while other studies such as Freed, So and Lazar (2003) and Llanes and Muñoz (2013) show no significant improvement. However, these apparently contradictory findings might be explained by other factors such as the length of stay in the host country or the instruments used to measure gains in L2 writing scores; for example, the studies by Serrano, Tragant and Llanes (2012) and Sasaki (2009) examined the writing development of participants spending one year or longer in the SA context whereas the studies by Freed, So and Lazar (2003), Llanes and Muñoz (2013) and Pérez-Vidal and Juan-Garau (2009) examined gains in L2 writing scores over a semester abroad.

Age, SA and L2 development
Most of the studies mentioned above have focused on university students, with the exception of Llanes and Muñoz (2013), which also included children. To our knowledge, only three studies have examined the effects of the SA learning context on the L2 development of adolescents. One of them is Lapkin, Hart and Swain (1995), who explored the global L2 improvement of a group of English-speaking adolescents engaging in a three-month SA in Quebec. The authors conclude that the SA was very positive since the post-test scores were higher than the pre-tests scores in all areas, and especially for aural skills. Several years later, Evans and Fisher (2005) set out to examine the impact of an SA experience on the four skills of a group of British adolescents engaging in an exchange programme (6-11 days) in France. The authors found that the exchange programme was positive and had greatest impact on listening comprehension and writing performance. Finally, a more recent study conducted by Llanes and Muñoz (2009) explored the effects of a short SA on the oral skills of a group of Catalan/Spanish students formed mostly of adolescents. The study examined several aspects of L2 oral fluency and listening skills and found that after their SA experience, the participants significantly improved their scores in 8 out of the 13 measures examined. The authors conclude that short SA experiences can have a positive impact on the participants' L2 oral and listening skills despite the large amount of first language (L1) use by the students.

SA and individual differences
A broad body of research has explored the effects of motivation on L2 development (Masgoret and Gardner 2003) but relatively few studies have examined the role of individual differences (including motivation) when learning occurs in an SA context. One of the few is Isabelli-García (2006). The author examined how motivation and attitude determine social interaction in the host country and found that motivation is not a fixed trait, but one that changes over time; attitudes towards the L2 and motivation are crucial in determining social interaction while abroad. Not surprisingly, highly motivated students and students with a more positive attitude towards the L2 experienced the greatest L2 gains. In a similar study, Hernández (2010) explored the effects of motivation and interaction on the development of L2 oral skills in participants studying abroad and AH. He found that SA participants improved their L2 oral skills significantly more than AH participants despite the similar patterns of motivation shown by both groups of participants. However, Hernández also found that SA students with higher integrative motivation interacted more than less motivated students, and that interaction in the L2 was a significant predictor of gains in L2 oral skills. In a more recent study, Llanes, Tragant and Serrano (2012) investigated the impact that individual differences such as motivation, attitudes and L2 contact have on the L2 written and oral performance of a group of students on an 'Erasmus' study-abroad programme. The authors found that those with higher linguistic expectations experienced greater gains, and that students studying abroad as part of a requirement by their home university experienced fewer gains than students who volunteered to go abroad. The authors also found that the more contact students had with people from their home country, the fewer gains they experienced.
Therefore, the research examining the effects of individual variables in an SA context seems to indicate that factors such as motivation, attitude and L2 practice are crucial for L2 development. However, again, all the studies reviewed here on individual differences have focused on undergraduate learners spending a semester abroad, while nothing is known about the effects (if any) that individual differences might have among younger participants (i.e. adolescents) spending less time abroad (three weeks).
The aim of the present study is to address the gaps identified above by exploring the impact of a set of individual differences on the L2 writing development of a group of adolescents studying abroad for a short period of time (three weeks). More specifically, the study seeks to answer the following research questions: 1. Does a short SA experience (three weeks) have a positive impact on the L2 writing skills of a group of teenage learners of English?
2a. Are the individual differences of initial predisposition, L2 use, perceptions and English knowledge prior to departure related to the development of L2 writing skills during this period?
2b. Do L2 learners who are more successful in developing L2 writing skills differ in terms of these individual differences from those who are less successful?

Participants
A total of 64 Catalan/Spanish bilingual learners of L2 English participated in the study. There were 18 males and 46 females. The participants were all adolescents (aged 12-17) engaged in a three-week SA programme in the south of England during the month of July 2011. They came from different schools all over Catalonia (Spain). Once abroad, they attended classes for three hours a day (in the mornings) from Monday to Friday. The four L2 skills were dealt with in these classes, although speaking and listening skills were practised more than reading and writing. Students were allocated to classes based on their L2 level and were mixed with other international students. All their instructors were native speakers (NSs) of English. The instructors were also in charge of afternoon activities such as sports and arts and crafts. In the evenings, students could take part in organised activities such as going to the disco or to a karaoke. All participants were accommodated in a dormitory. They were encouraged to mix with other international students at meal times and at weekends, although no specific weekend activities were programmed.

Instruments and procedure
This study has a pre-test and post-test design. The pre-test was administered on the second day of the participants' arrival in the host country, while the post-test was administered two days before their return to Catalonia. The time elapsing between the pre-test and the post-test was 18 days, and in order to minimise any possible task-repetition effects, participants at the pre-test were not informed that there would be a post-test.
Participants took several tests, but for the purpose of the present paper, only the writing task will be considered (the participants' results in the other tasks will be examined in other papers). Participants were shown a short comic strip of 10 frames with all speech bubbles blanked and were asked to write the story depicted in the strip (see Appendix 1). They had 10 minutes to complete this task and were encouraged to write as much as possible within the given time. They were then asked to fill out a questionnaire that elicited some biodata and assessed their initial predisposition, motivation and attitude towards the L2, and amount and type of L2 practice during their stay abroad. Table 1 shows the key factors of interest while an abbreviated form of the questionnaire appears in Appendix 2. The questionnaire was administered in the participants' L1 and they took as much time as they needed to fill it out. Given that we do not know what their initial L2 level was, the term 'gains in L2 writing scores' will be used to refer to any language changes from the pre-test to the post-test.

Measures
The four L2 writing areas examined were fluency, lexical richness, grammatical complexity and accuracy. Written fluency was computed through the total number of words (tokens) written (TOK). This measure was chosen because it had been previously used with similar participants and was found to discriminate well (Muñoz, Tragant, and Torras 2010). Lexical richness was calculated by dividing the number of types by the square root of the number of tokens (Guiraud's Index [GUI]). This measure was chosen again because it had proved effective in earlier studies (Llanes and Muñoz 2013). Two measures accounted for grammatical complexity: the percentage of types of verb forms out of the total words (%VF) (Muñoz, Tragant and Torras 2010) and the number of clauses divided by the number of T-units (C/T) (Wolfe-Quintero, Inagaki and Kim 1998). Hunt (1965: 20) defines a T-unit as 'one main clause with all subordinate clauses attached to it'. Finally, accuracy was measured through the percentage of correct verb forms out of the total number of verbs used (%CV).

Results
Preliminary assumption testing was conducted to check for normality and it was found that three out of the five measures violated the assumption of normality according to the Kolmogorov-Smirnov test: TOK (fluency), C/T (grammatical complexity) and %CV (accuracy). In order to answer the first research question, which asked whether a three-week SA experience promoted gains in the L2 writing scores of a group of teenage learners of English, a paired-samples t-test was run for the lexical richness measure (GUI) and the grammatical complexity measure (%VF), whereas a non-parametric Wilcoxon Signed Ranks Test was run for the three measures that were not normally distributed (TOK, C/T and %CV) to see whether there were any significant differences between the pre-and post-test scores. As can be seen from Table 2, the participants scored higher in the post-test on all the measures. However, not all these improvements turned out to be significant. Improvement was significant for TOK (Z = −4.583, p = .000, d = .445), GUI (t(60) = −4.947, p = .000, d = .420), %VF (t(60) = −1.648, p = .012, d = .09), and %CV (Z = −2.091, p = .037, d = .14). For all these significant measures, effects sizes were medium except for %VF and %CV, for which the effect size was small (Cohen 1988). Thus, the results from the t-tests and the Wilcoxon tests indicate that participants experienced a significant improvement in four out of the five measures after spending three weeks abroad.
Research question 2a focused on whether individual differences were related to gains in L2 writing scores, and in order to answer it, bivariate Pearson correlations were run between the students' gains in each measure (calculated by subtracting the pre-test scores from the post-test scores) and the results on the different items of the questionnaire. As shown in Table 3, only three statistically significant correlations were found. Motivation (Question 12a) was significantly and positively correlated with %CV, and as was perceived Amount of English learned (Question 16). There was a significant negative correlation between the students' English mark at school and %VF (complexity).  In order to answer research question 2b, which asked if more successful L2 learners differed from less successful L2 learners in terms of individual differences, two groups of participants were identifiedmore successful and less successful L2 learnersbased on a score derived from all five writing measures. Since the scores for each of the measures were of a different type using different scales, they had to be transformed into z-scores in order to standardise them (Dörnyei and Schmidt 2001;Strube, Van de Craats and Van Hout 2012). The z-scores for each of the five measures were then summed to give a final z-score, Those participants with a positive z-score (z-score > 0) were considered more successful learners (n = 27), while participants with a negative z-score (z-score < 0) were considered less successful learners (n = 33). Four participants were excluded because it was not possible to calculate a final z-score as they were missing one of the five measures. Next, bivariate correlations between the L2 gains and the individual difference variables were run for each of the groups separately. For the group of more successful language learners, several variables were found to be correlated with their L2 gains.
As shown in Table 4, fluency was positively correlated with Motivation to enrol in the SA programme (Question 9) while Lexical richness (GUI) was significantly and positively correlated with having taken previous intensive courses AH (Question 10c). Grammatical complexity (C/T) was significantly and positively correlated with having attended previous summer camps in English (Question 10a). The grammatical complexity measure %VF was the one most affected by individual variables: those participants who perceived they had learned more English during their SA (Question 16) were the ones to experience greater gains in this measure. Curiously, the easier participants perceived the English language (Question 13), the fewer gains in %VF they experienced. Speaking English either with people from other countries or with English NSs (Question 25) was also significantly and positively correlated with %VF and Use of English during meal times and Use of English in accommodation (Questions 26b and 26d respectively) were both positively correlated with grammatical complexity. Finally, Motivation to learn English (Question 12a) was significantly but negatively correlated with accuracy (%CV). In other words, more motivated participants demonstrated fewer gains in accuracy.
Correlations were then run for the group of less successful language learners. As shown in Table 5, only two significant correlations were found: the reported FL Use during SA weekends and free time and weekends was negatively correlated with grammatical complexity (%VF), while reported FL Use during SA organised activities was positively correlated with accuracy (%CV).
Thus, in terms of correlations, results show that for the group of more successful learners, more variables come into play than for the less successful learners. In order to explore further the differences between these two groups, the questionnaire responses regarding other aspects of their SA experience were also examined and compared. The detailed results for more successful and less successful groups are given in Appendix 3. In what follows, we comment on areas where the data suggested possible differences between the groups.
The responses to Question 7 indicate that while only a quarter of the less successful learners had attended CLIL programmes (programmes in which a subject is taught through a foreign language (FL) with both the aim of learning the content and the FL), over a third (37%) of the more successful learners had. Although more successful and less successful participants did not differ much in terms of previous experiences of SA programmes and summer L2 courses AH (Questions 10 b and c), around two thirds of the more successful participants reported having attended an English language summer camps at least once, compared with just under half the less successful participants (Question 10a). Another difference between both groups of participants was their motivation to learn English (Question 12). Surprisingly, less successful learners reported greater motivation: almost 70% reported liking English 'a lot' or 'very much indeed' compared with 52% for the more successful group. Another finding was that less successful learners perceived English as more difficult to learn than their more successful peers (Question 13). When asked about the advantages of staying in dormitory accommodation (Question 15a), more successful learners tended to mention that it was a more social experience (37% compared with 21% of the less successful group). When asked about the disadvantages, more of the more successful learners mentioned concern about the extent to which the L2 would be used (37% compared with 12.1%), while less successful learners reported other reasons.
A surprising difference was that, in general, the more successful learners felt they had learned less during their study abroad than their less successful counterparts: on Question 16, over half the more successful group (56%) responded that they had learned 'not much' or 'a bit' versus only 15% of the less successful group. In the same vein, around 40% of the successful group evaluated the English classes in the SA context somewhat negatively (responding 'not really' or 'a bit' to Question 23 which asked if they liked the English classes) while only 9% of the less successful learners were similarly negative. It should be noted however that a high percentage of both groups (40% and 45% respectively) failed to respond to this question.
On the other hand, more successful learners reported willingness to enrol again on an SA programme: on Question 20, almost two thirds (63%) said they would 'probably' or 'definitely' enrol again, compared with just under half (48%) of the less successful group. The more successful group also reported less homesickness: when asked if they missed family or friends (Question 21), almost half (48%) the successful group reported 'not really' or 'a bit' compared with 30% of the less successful group. Twenty-four per cent of the less successful group said they missed family and friends 'quite a bit', 'a lot' or 'very much indeed', compared with 14% of the more successful group. However, there was again a high proportion of non-response: 37% for more successful group and 46% for the less successful group on both Question 20 and 21 so these findings need to be considered as tentative only.

Discussion
The results concerning the first research question, which asked whether a three-week SA experience had any positive influence on the L2 writing skills of a group of teenage learners of L2 English, revealed that participants improved significantly in four out of the five measures of L2 writing used in this study. This finding is in line with the beneficial effects of SA found for areas such as oral skills (Freed 1995;Llanes and Muñoz 2009), vocabulary (Dewey 2008;Foster 2009) and sociolinguistic skills (Regan, Howard and Lemée 2009). It also confirms the positive impact on writing found in some previous research (Sasaki 2009;Serrano, Tragant and Llanes 2012), but runs counter to some studies that did not find an advantage for the SA context (Freed, So and Lazar 2003;Llanes and Muñoz 2013). The reason for the positive outcomes of the present study could be the participants' age (teenagers as opposed to adults and children) or the measures used, which may be more susceptible to revealing L2 writing gains.
In relation to the influence of individual differences on gains in L2 writing scores (research question 2a), few significant correlations were found, and those which were found were somewhat surprising. Those reporting higher scores on Motivation to learn English and perceived Amount of English learned showed fewer gains in writing accuracy (%CV) while those with higher marks in English at school displayed fewer gains in grammatical complexity as measured by %VF. It could be the case that participants with a higher mark in English focused on other aspects rather than grammatical complexity and that they improved other areas at the expense of grammatical complexity.
This study also explored whether individual differences could explain the success (or lack thereof) in L2 learning. It was found that for more successful language learners, more variables came into play than for less successful learners. Although the correlations between L2 gains and the constructs investigated in the questionnaire did not reveal important differences between the two groups, the more detailed consideration of the questionnaire data did suggest some possible connections. It was found, for example, that for the more successful group, the amount of L2 contact seemed to be a key factor influencing gains in grammatical complexity, as measured by %VF, a finding that confirms the role of practice (DeKeyser 2003;.
A surprising finding was that less successful learners perceived they had learned more English than the successful learners. These findings may indicate that the more successful learners tended to expect more of themselves. Rather more predictable findings were that more successful learners tended to have had greater exposure to English through CLIL classes and L2 summer camps, and experienced less homesickness. They also judged learning English less difficult than the less successful learners.

Conclusion and further research
This study found that even a short three-week SA experience had a positive impact on the L2 writing skills of a group of adolescent learners and that these improvements appeared to be influenced by certain individual factors: more successful and less successful learners showed some different motivational and attitudinal patterns that might account for differences in outcome. However, these results must be taken with caution given the limitations of the present study. First, the data gathered through the questionnaire were self-reported which can undermine their validity. Second, our answers to research question 2b are exploratory and descriptive, based on responses from relatively few students with, for some questionnaire items, a high proportion of non responses, so these results are not generalisable and can only be interpreted as a first step to a better understanding of possible individual variation between students. Another limitation is that written development was assessed through a very specific task; furthermore, the same picture story was used at pre-test and at post-test. It could therefore be that task repetition could account for some of the L2 writing gains, though this risk was mitigated by not informing participants that the picture story would be used again at posttest. Further research might explore whether different types of writing task show similar results. It is also important that the effects of SA on this under-investigated age group be explored across a wider range of L2 skills and knowledge. Nevertheless, despite its limitations, the present study does confirm the value of short SA stays for adolescents, but also highlights that a number of individual factors may influence participants' linguistic gains. Programme coordinators might therefore wish to take account of some of these factorssuch as prior exposure to L2 in CLIL classes and summer camps, the possible impact of homesickness and L2 practicein the design of L2 curricula incorporating SA experiences.
hardly ever a few times sometimes almost always always During your stay in general, which language did you use the most? (Choose one option): Much more Eng than Cat/Sp/ A bit more Eng than Cat/Sp/ As much Eng as Cat/Sp/ A bit more Cat/Sp than Eng/ Much more Cat/Sp than Eng