Erasmus students using English as a lingua franca: does study abroad in a non-English-speaking country improve L2 English?

There is a lack of research on the impact of study abroad (SA) on the development of L2 English when students study in non-anglophone countries. The aim of the present study is to fill this gap by examining 39 Catalan/Spanish students who, as part of an Erasmus exchange, spent a term at universities in non-English-speaking European countries. In this context, English was used as the vehicular language for their studies and in their daily extracurricular activities when interacting with other students. Our research focuses the impact of this specific type of SA on students’ L2 English proficiency, and in particular their writing skills. Before and after the SA, students completed the Quick Oxford Placement Test (a general measure of L2 proficiency) and also drafted a short written paragraph in English. Their writing was analysed for syntactic complexity, lexical complexity and subordination. The results show that participants improved significantly after their SA on two out of the four measures: general L2 proficiency and lexical complexity. Though threshold levels of general proficiency have been posited for students’ ability to benefit linguistically from SA, in this case, L2 proficiency at the outset of the SA experience was found not to influence the development of writing skills, except on the measure of subordination. Implications are drawn for further research and pedagogic practice in developing English as a lingua franca skills in a European context.


Introduction
As a result of internationalisation (Blumenthal and Laughlin 2009), the popularity of studyabroad (SA) programmes has increased in tandem with the need for graduates to function in global working environments. This is no surprise: the SA context is seen as one of the most effective contexts for learning a second/foreign language (L2) in that it offers what seems to be an ideal combination of formal (classroom) and informal (out-of-class) L2 learning (Collentine 2009;Freed 1995). One of the key advantages is assumed to be the authentic, unlimited and potentially 'rich' nature of L2 input that students are exposed to, and research has found that study in a country where another language is spoken is generally positive for learning that language, particularly for the development of oral skills (Collentine 2009;Llanes 2011).
However, to our knowledge, few studies have investigated the effects of SA on the development of L2 English in countries where English may be used as a lingua franca in academic contexts but where it is not the official language of the country (e.g. Spanish students with L2 English studying in The Netherlands or Germany). Kalocsai (2014) is one of the few researchers who has looked at a similar context.
The lack of research on this type of SA is remarkable given the popularity of student exchanges across Europe under the Erasmus programme. The latter programme was set up (and funded) by the European Union over 25 years ago in order to promote mobility between European universities as a pathway for greater cohesion among European citizens. It has been highly successful in promoting student exchanges between European countries, such that today a large number of Spanish students, for example, enrol in SA programmes in countries such as Germany, Bulgaria and Poland (see the data from the European Commission, available from http://ec.europa.eu/education/tools/statistics_en.htm#erasmus). They do not necessarily learn the official language of their host country as L2 English typically functions as a lingua franca. But what impact does this use of English, away from assumed contact with 'English native speakers', have on the development of L2 English skills? The present paper 1 sets out to explore the development of L2 English writing competence among a group of Catalan/Spanish learners of English after SA at universities in non-English speaking European countries.

Background
Although popular wisdom portrays SA as the ideal context to make progress in a second language because of the combination of formal and informal opportunities for language learning and practice (Allen 2010;Davidson 2007;DeKeyser 2007), research (Llanes 2011) has raised questions about how specifically SA should be incorporated in institutional policies and programme design, and in particular, which L2 domains are likely to benefit most from SA. SLA research has distinguished three contexts where L2 learning takes place (Collentine 2009; Collentine and Freed 2004). The foreign-language instructed (FI) setting or at home (AH) context refers to L1 local settings, where students learn the L2 within L1 educational settings, typically for around 2-4 hours a week. The intensive domestic immersion setting is where language programmes in the L1 setting offer more intensive contact with the L2 (i.e. 25 hours a week or more) (Serrano, Llanes and Tragant 2011). Such an immersion setting obviously provides students with more opportunities to use the L2 for functional purposes. Finally, the SA setting refers to contexts where the L2 is institutionally, socially and functionally implemented. All three contexts are compared in the section below which reviews the research on language gains in SA contexts.

Language gains
The L2 domains found to develop most after SA are oral fluency, vocabulary acquisition and sociopragmatic skills. Oral fluency consistently shows greater gains after SA than other skills. Llanes and Muñoz (2009) studied oral proficiency gains by Spanish students of English after short stays abroad and found that students generally improved on most of the oral fluency measures, including gains in accuracy, though this has not been extensively evidenced. Likewise, Segalowitz and Freed (2004) examined participants' gains in oral fluency and general proficiency, and whether the measures of oral performance were related to L2-specific cognitive measures of lexical access, among others. The study compared US students learning Spanish in an AH setting and in a SA setting, and found that SA students achieved more language gains on cognitive measures (lexical access and attention control) and oral proficiency measures than the AH students.
There are also abundant studies on the acquisition of vocabulary in SA settings, with positive results for students' lexical improvement, as evidenced, for example, by lexical complexity measures or receptive vocabulary tests (Dewey 2008;Serrano, Llanes and Tragant 2011). The likely explanation for the consistent evidence of vocabulary improvement no doubt has to do with the higher number of hours of SA classroom instruction, and a more extensive exposure to language outside the classroom in the SA context (Dewey 2008).
Regarding the development of sociopragmatic competence in an SA context, Regan (1995) reported an increase in the deletion of the French adverb ne ('no, not') by Irish students of French during their SA in France, which made the students' usage consistent with native speaker standards, even if some overuse was also observed. Marriott (1995), however, found that the development of politeness strategies by Australian students on SA in Japan differed on an individual basis, and that after one year abroad, their oral performance still deviated significantly from Japanese conventions. Siegal (1995) also examined politeness strategies by US students on SA in Japan. Her qualitative study provides evidence to account for individual differences in the development of Japanese politeness strategies, which seems to be strongly influenced by each student's wish either to adopt the different linguistic politeness conventions of Japanese or to maintain their L1 conventions in the SA context.
Listening comprehension after SA has not often been a key focus of research, but the few results that are available suggest that SA is beneficial for the improvement of oral comprehension. Llanes and Muñoz (2009), for example, examined listening development by participants over a short SA period (3-4 weeks) and found that they improved significantly on the listening comprehension post-test. Similarly, Rodrigo (2011) found that a 5-week overseas experience resulted in gains equivalent to a semester-long AH course. However, contrary to Llanes and Muñoz (2009) and Rodrigo (2011), Cubillos, Chieffo andFan (2008) compared listening comprehension gains of SA and AH students in the USA and found that both groups of students experienced similar gains. Reading skills have also received little attention. As with listening, some of the few available studies have shown positive results as a result of SA (Davidson 2010) but there are also studies that have found no advantage for SA over AH or immersion (Dewey 2004).
The L2 domains which show less clear-cut linguistic development after SA are morphosyntax, phonology and writing. For example, DeKeyser (1991) found that SA students did not obtain better grammatical results than their AH counterparts. Likewise, Collentine (2004) measured SA students' acquisition of grammatical features in Spanish and found that they did not make as much progress as the AH students on these features (e.g. verbs and connectors) which had been the focus of formal instruction. As for phonology, studies on L2 speech learning and phonology acquisition in an SA setting are scarce and somewhat contradictory. Diaz-Campos (2004) examined the effects of learning context (SA vs. AH) on the pronunciation of English learners of Spanish, focusing on the accurate pronunciation of consonants. The results show a similar development in the production of consonants by students in the two learning contexts. Similarly, Mora (2008) explored the phonological competence gains by Spanish FI students compared with Spanish SA students of English and found that FI students showed clearer gains in perceptual ability and in some phonemic contrasts than the SA students.
In relation to studies looking at the development of writing after SA, findings have been inconclusive and there has been controversy concerning methodology. Batardière's (1993) study of Irish Erasmus students found some progress in basic oral-aural skills but no perceptible gains in writing after SA. This study was based on self-reported data and was criticised for its lack of reliability (DeKeyser 2007) and also because the participants had not been trained to identify language gains in writing (Pérez-Vidal and Barquin 2014). However, even using a range of more objective measures, Freed, So and Lazar (2003) found no evidence that setting had any impact when they compared the writing of US university learners of French after SA vs. after AH study. The measures used in their study included syntactic complexity, accuracy and written fluency, as well as overall evaluation of written fluency by native-speaker judges. Sasaki (2004Sasaki ( , 2007Sasaki ( , 2009Sasaki ( , 2011, Pérez-Vidal and Juan-Garau (2009), and Pérez-Vidal and Barquin (2014) have also had inconsistent results from their studies on SA and writing. Sasaki observed Japanese university students learning English in SA and AH contexts over a period of three and a half years. Quantitative data (composition scores) and qualitative data (interviews) were collected. Sasaki found that both SA and AH groups showed improvement on overall L2 proficiency, composition quality and writing fluency. However, the SA group showed greater development in measures such as writing strategies and writing styles, no doubt due to their greater gains in vocabulary over the period. Pérez-Vidal and Juan-Garau (2009) analysed writing produced by Spanish students of English in an SA context and also found gains for the SA setting over the AH setting. SA students showed greater gains in lexical complexity and written fluency, though not in accuracy.
Finally, Pérez-Vidal and Barquin's longitudinal study (2014) examined the writing development of 73 Erasmus students of English at four data collection times over a two and a half year period. Their written production was compared to that of native speakers and was assessed for measures of complexity, accuracy and fluency. Their results showed that a 3-month period of SA led to significant gains in writing for fluency, accuracy and lexical complexity, but no gains were observed after a comparable period of FI instruction. These later studies, unlike those of Batardiére (1993) and Freed, So and Lazar (2003), suggest that the SA learning context may be associated with language gains in writing.

Threshold levels
Results are also inconsistent regarding possible threshold levels in L2 proficiency that learners may need to attain before they can benefit linguistically from SA. Llanes and Muñoz (2009) studied whether short stays abroad in an anglophone country resulted in language improvements for Spanish students of L2 English. The results revealed that participants with a lower starting level experienced greater language gains than more proficient students. Lapkin, Hart and Swain (1995) also found that students with a lower pre-departure language level showed greater linguistic gains after SA, especially for listening and speaking skills, in line with previous research (Brecht and Davidson 1991;Brecht and Robinson 1995;Freed 1990).
However, Brecht and Robinson (1995) found that the best predictors of overall proficiency gains in the SA context were high grammar and reading pre-test scores. Likewise, DeKeyser's (2010) mixed-methods research showed that L2 knowledge prior to SA was that the best predictor of more successful performance after SA; in other words, students whose grammar knowledge was better in pre-tests did better on speaking tests after the SA experience. DeKeyser's qualitative findings also showed that those students with a stronger L2 declarative knowledge were better able to monitor themselves during their SA stay.
Taking a cognitive approach, Segalowitz and Freed (2004) concluded that an initial threshold level of basic word recognition and lexical access processing abilities may be necessary for oral skills to develop. The idea of a cognitive threshold is further informed by evidence from more advanced learners. Hulstijn and Bossers (1992) found that more advanced students had developed a larger working memory capacity, because they had partly automatised lexical retrieval processes. This capacity may be significant in L2 language development because it 'allows learners to process longer segments of input and hold longer strings in their heads for incipient output' (Payne and Whitney 2002;in Lafford and Collentine 2006: 117).
In summary, there is an increasing general agreement among researchers that SA learners must first reach a threshold level of L2 proficiency in order to benefit from SA, From a practical perspective, researchers have called for more research on this issue (Collentine 2009; Llanes 2011) so as to be able to inform institutional policies and programme designers in SA about how to optimally prepare students for the SA experience.

Research questions
Most research on the impact of SA looks at L2 development in a context where that L2 is officially used and is the most socially present language. However, students on the Erasmus programme may choose to undertake SA in non-English speaking European countries such as Norway, Poland, Belgium, Austria or Greece. In these countries, English tends to be used in the academic environment as a lingua franca. In such contexts: (a) Erasmus students tend to socialise with other international students rather than with the local population and (b) they perform academic work in English, which means that their use of English can be characterised as English for academic purposes (EAP), involving essay writing, oral presentations, listening to lectures, and other academic skills. The research questions for this study are framed as follows: (1) What is the effect of an SA period in a non-English speaking country where English is used as a means of communication on L2 English development measured by a general proficiency test and various assessments of writing? (2) What role does initial L2 English level play in L2 English development in this context?

Method Participants
Participants in this study were university students from two universities in Catalonia, Spain: the University of Lleida (UoL), which offers a wide range of degrees in different areas of study, and the Polytechnic University of Catalonia, a large metropolitan university specialised in engineering and technology. As part of an Erasmus exchange programme, the students participated in SA of approximately 15 weeks at a university in a country where English was not the official language, but was used as a lingua franca. This took place in either the autumn term or the spring term of the academic year. A total of 39 students aged 19-33 (M = 22.19) were selected for this study: 22 from the UoL and 17 from the Polytechnic University of Catalonia at Vilanova i la Geltrú (UPC).
The UPC students were all majoring in engineering (industrial design, mechanical, electronics, electrical and computing), while UoL students were majoring in a variety of degrees from humanities, social sciences, health sciences, and science and technology. For the purposes of this study, the participants were considered homogeneous in terms of language background (Catalan/Spanish), age and university experience; neither the home university nor the students' degree discipline was considered to have an influence in the design of this study. The students had varied levels of language proficiency in L2 English before SA and this variable was a key consideration in our study.

Procedure
Participants were selected in collaboration with the international office responsible for student mobility at each university. All completed the same test, comprising language components and a questionnaire on intercultural attitudes and European citizenship. The test was preceded by a demographics section inquiring about age, language background, attitudes towards the SA, and previous language learning and SA experience. The participants were asked to give their consent to the confidential use of these data for this research. The test was administered twice, first as a pre-test a few months before departure for SA and then as a post-test on return. 2 The language tests relevant to this study included a general language proficiency test (the Quick Oxford Placement Test, OPT), consisting of a battery of multiple-choice questions focusing on the accurate selection of grammatical and lexical items. To measure writing skills, students were given 15 minutes to produce a short piece of writing (150-200 words) in English on a general topic, 'The advantages or disadvantages of studying abroad, according to your own personal experience (if any) and/or beliefs'. This topic was considered relevant to the participants' situation; they could be assumed to have something to say about it and it did not advantage students from any particular academic discipline.
Following Wolfe-Quintero, Inagaki and Kim (1998), the participants' writing was assessed according to the following variables: syntactic complexity, lexical complexity and subordination. Syntactic complexity was measured in terms of number of words per T-unit (WDS/TU). 3 Lexical complexity was measured by means of Guiraud's index of lexical richness (GUI), which corresponds to the number of types divided by the square root of the total number of tokens. Subordination was measured as number of clauses per T-unit (CL/TU). 4

Data analysis
The data were transcribed and coded by the three authors using CLAN (MacWhinney 2000). Interrater reliability was calculated by comparing the codings of 15% of the data. The comparison was mostly based on the number of T-units and of clauses given that the CLAN automatically calculated the number of types and tokens that were used to compute the syntactic complexity and Guiraud's measures. There was 97.3% coding agreement and the few cases where there was disagreement were discussed until an agreement was reached.

Results
Descriptive data for the four linguistic tests are shown in Table 1. Preliminary assumption testing of the pre-test scores was conducted to check for normality and it was found that three out of the four measures examined -OPT (placement test), WDS/TU (syntactic complexity) and CL/TU (subordination)violated the assumption of normality according to the Kolmogorov-Smirnov test. Therefore, in order to answer Research Question 1, which asked whether participants going abroad to a non-English speaking country improved their L2 English, a Wilcoxon signed-rank test was run for OPT, WDS/TU and CL/TU, and a Paired Sample t-test was employed for the remaining measure, the GUI (lexical complexity).
The For the two significant measures (OPT and GUI), the effect sizes were small and medium respectively, indicating that improvement on the OPT was not uniform, but on the GUI, it was fairly uniform. Thus the data showed that after their SA experience, the participants somewhat improved their general L2 English proficiency and increased their lexical complexity, but wrote shorter and less complex sentences.
In order to answer Research Question 2, which asked whether initial L2 English proficiency level plays a role in the development of L2 English after SA in the English as a lingua franca (ELF) context, bivariate correlations were performed between the participants' scores on the OPT in the pre-test and the gains they experienced in each of the other three measures (syntactic complexity, lexical complexity and subordination). As shown in Table 2, only one significant correlation was found, namely between the participants' initial L2 proficiency and CL/TU [p = .029]. In other words, participants with a higher L2 English level at the start of their SA experience produced significantly more complex sentences (i.e. with more subordination) when they returned. In order to ascertain what proportion of these gains could be explained by initial proficiency level, a linear regression analysis was conducted with OPT pre-test scores as the independent variable and CL/TU as the dependent variable. As shown in Table 3, 35.4% of the gains in CL/TU was explained by initial L2 proficiency.

Discussion
Unlike previous studies, our study focused on L2 English development in a SA context where English was used as a lingua franca, not as an official language. One might predict that this context would offer students a different quality and quantity of L2 English input and interaction from one where English was the official language. It is difficult in the present study to specify exactly how each of the participants used L2 English during their SA. They came from a variety of academic degree progammes and the activities they undertook during their SA varied from completing their final thesis to attending courses and engaging in collaborative project work with other students and tutors. The type of instruction received at the host university may also have influenced students' results, especially considering that this study focuses on the development of writing skills. Some students may have been required to produce formal written assignments or may have received formal instruction in writing, while others may have not. The results reported above indicate that there was indeed significant improvement in overall English language proficiency, according to OPT scores, which is in line with research that has found overall language proficiency improvement as a result of SA (Brecht and Robinson 1995;Lapkin, Hart and Swain 1995). This result points to the beneficial nature of SA on L2 English even if this SA takes place in non-English speaking countries. However, detailed analysis of the OPT results also showed that proficiency gains were not uniform across students. This might be explained by students' varied levels of English proficiency prior to SA as well as by the potential diversity of experience at the host university, as mentioned above.
Another significant result from this study is the evidence of improvement in L2 English lexical complexity. This is in line with DeKeyser's (1991) study and suggests that SA provides plenty of opportunities for recurrent focus on meaning, which in turn potentially allows for focus on form, and this appears to be the case even in contexts where English is used as a lingua franca rather than as the official language of the country. It is likely that the L2 English language practice resulting from participation in academic and social activities during the SA contributed to the development of English lexical complexity, a finding that echoes studies by Foster (2009) andIsabelli-García (2006).
On the other hand, no significant gains were found in written syntactic complexity and subordination, with students obtaining even lower scores in the post-test. In other words, they wrote shorter and less complex sentences after SA than they did before. This could be explained by the fact that they were not necessarily required to practise their L2 English writing skills during the SA, nor did they necessarily receive any formal L2 English instruction. They no doubt used oral English regularly during their stay and possibly read more in L2 English (with a resulting increase in particular in their lexical complexity). Assuming they gained in confidence in using their L2 English skills, they may have approached the writing post-test with greater confidence than the pre-test, focusing more on content than on form and structure. In response to Research Question 2, participants' initial level of L2 English proficiency was found not to influence significantly the L2 writing gains they made during their SA, at least in the areas of syntactic complexity and lexical complexity. However, gains in subordination in writing were related to initial proficiency level, such that students with a higher initial L2 English level tended to write more complex sentences in the post-test. This result may be attributable to the fact that students with a higher proficiency level in L2 English prior to their SA may have in effect received more formal L2 English instruction over the years, and this may have given them a wider range of linguistic resources to help them produce more sophisticated writing. Another tentative explanation could be that participants with a higher level of proficiency need not allocate so much attention to areas such as vocabulary, for example, which in turn might have allowed them to allocate more attention to other aspects such as complexity.
This study thus found evidence of linguistic gains in L2 English after SA even in a context where English was a lingua franca rather than the official language of the country. It should be noted that these linguistic gains were identified in written tasks, rather than in speaking, where 'standard' SA has typically been found to have a major impact (Llanes 2011).

Conclusion and further research
This article has explored the development of English (L2) after an SA experience in a non-English speaking destination, but where English was used as a lingua franca. It was found that spending a term abroad in an academic context in a non-English speaking country resulted in significant gains in two out of the four language measures examined, namely general English proficiency and written lexical complexity. These findings are considered positive, particularly if we take into account that several studies that have examined writing skills as a result of 'standard' SA, that is, in a country where the L2 is the official language, showed no gains at all.
However, a significant limitation of this study is that the participants came from different academic backgrounds and their SA experiences took place in very different non-English-speaking countries The amount and quality of ELF used will vary from context to context; use of ELF in Denmark may well be very different from use of ELF in Poland or Italy, for example. It is thus important to conduct further research to identify differences in ELF usage in different contexts and to explore whether there are any significant differences in students' gains depending on the non-English speaking country where they complete their SA (e.g. Nordic countries vs. Mediterranean countries). Another limitation of the present study is that only gains in written measures were examined, when it is known that 'standard' SA experiences have been found to be especially beneficial for the improvement of oral skills. Thus, further research should look at any gains in oral skills of participants going to a non-English speaking destination. Yet another limitation is the fact that the pre-test was administered some time before the participants departed for their SA, and it may be that for at least some participants, their L2 English improved between pre-test and departure. Furthermore, the topic of the writing test might have generated a more enthusiastic response from students after they had returned from SA than before leaving and this could conceivably have skewed results in favour of the post-test. Indeed, results on a paired-sample t-test showed that the participants wrote significantly more words in the post-test than in the pre-test [t(37) = −2.027, p = .05].
Despite these limitations, the present study fills an important gap in the field of SLA and has significant pedagogical implications. The language gains observed in academic settings where English is used as a lingua franca underscore the importance of developing teaching materials focusing on ELF and, more specifically, on EAP. In relation to the latter, further attention may need to be paid to the language/writing instruction that students receive during their SA, such as the development of teaching materials focusing on academic writing for ELF settings. Further, if we understand better the kinds of interactions and activities in which students engage in ELF during their SA in non-English-speaking countries, we may be able to promote course materials based on ELF interaction in authentic academic settings, which could help students reflect on, and prepare for, interaction in international environments. Taking a similar approach to resources developed in the USA by Feak, Reinhart and Rohlck (2009), such materials could focus specifically on realistic ELF situations in Europe.