The use of language learning strategies in a second and third language : The case of foreign language majors

Although multilingualism has become a fact of life in the last few decades, this phenomenon has largely failed to find a reflection in research on language learning strategies. Even when scholars have addressed this issue, it has mostly been done with the purpose of proving the advantage of multilingualism over bilingualism, and scant attention has been given to how the nature, utility or status of a particular additional language can impact the frequency and patterns of strategy use. The present paper seeks to partially fill this gap by investigating the employment of strategies by 107 Polish university students majoring in English and, at the same time, being required to reach a high level of proficiency in another additional language. The data were collected by means of the Strategy Inventory for Language Learning (Oxford, 1990) and interviews conducted with selected participants. A combination of quantitative and qualitative analysis demonstrated that strategy use in the second language was higher than in the third language, both overall and with respect to specific groups of strategies, mostly traditional and memory strategies were deployed, and the outcomes could be attributed to the proficiency level in both languages and varied motivation to master these languages.


Introduction
While it is certainly possible to find individuals who are monolingual in some societies, it is also true that the world is currently becoming more and more multilingual, with the phenomenon of multilingualism, or "coexistence, contact and interaction of different languages at the societal or individual level" (Wei, 2013, p. 26) becoming gradually ubiquitous, simultaneously impacting and being impacted by changes happening in different spheres of life (cf.Aronin, 2015;Zarobe & Zarobe, 2015).On the one hand, multilingualism can be seen as a natural order of things resulting from the coexistence of social groups speaking different first languages, prolonged contacts with immigrant communities, or simply the rampant process of globalization which has been made possible, among other things, by access to cutting-edge technologies such as the Internet or the influential social media applications.On the other hand, the ability to speak several languages is also actively promoted by numerous countries which have been enacting laws making it possible for instruction in two or more foreign languages to be initiated early in the process of education.In effect, Aronin and Singleton (2012, p. 1) point out that "the development of multilingualism in the world has reached a critical point in terms of scale and significance," while Cenoz (2013, p. 71) explains that "differences between second language acquisition (SLA) and TLA [which can by and large be equated with the concept of multilingualism] have been neglected in SLA [second language acquisition] research and in studies on bilingualism."Such developments, however, have on the whole failed to find a reflection in research on language learning strategies (LLS) that has primarily focused on the application of strategic devices in a second or foreign language (L2), giving very little attention to the differences in this respect between various additional languages that learners may aspire to learn in succession, as well as the possible transfer of strategies between different languages (see e.g., Cohen, 2011;Griffiths, 2018;Oxford, 2011Oxford, , 2017;;Pawlak, 2011, for an overview of existing research on language learning strategies).The present paper is intended to rectify this problem by reporting the results of a study which was conducted with the purpose of comparing the use of LLS in a second (L2) or third (L3) language, reported by students who were majoring in English, but were required at the same time to achieve a high level of mastery in yet another foreign language.The first part of the paper will be devoted to a succinct overview of the scant body of previous research that has tapped into the use of strategies by multilingual learners.The second part will be devoted to the presentation of the research questions, the design of the study, its findings and the discussion of these findings.The paper will close with the consideration of future directions of research on the use of LLS in different additional languages and pedagogical implications that could enhance the effectiveness of everyday teaching of foreign languages, irrespective of the order in which they are learnt.

Previous research into the use of LLS in L2 and L3 (L4) language acquisition
When one examines state-of-the-art publications dealing with language learning strategies (e.g., Cohen, 2011;Griffiths, 2018;Griffiths & Cansiz, 2015;Griffiths & Oxford, 2014;Oxford, 2011Oxford, , 2017;;Pawlak, 2011), it immediately becomes clear that strategies are typically considered with respect to L2 and only very infrequently are any other additional languages brought into the picture.In fact, when the use of LLS is considered in such cases, it typically involves comparisons between learners of one and more additional languages, the conceptualizations of LLS are sometimes fuzzy and imprecise, to say the least, or simply the way in which the findings are interpreted, also, by scholars citing these studies, tends to focus on the comparison between bilingualism and multilingualism rather than differences between strategy use in L2 and L3.
In one of the first studies exploring the use of LLS by multilinguals, Miβler (2000) used the German version of the Strategy Inventory for Language Learning (SILL, Oxford, 1990) as well as interviews to determine whether the number of languages known by 125 university students had an effect on LLS use.She found that greater experience in language learning was indeed accompanied by more frequent application of strategies, but this effect was mediated as well by individual difference variables.In a later study, Kemp (2007) employed a 40-item questionnaire based on a 5-point Likert scale in which participants could also include their own comments to explore the application of grammar learning strategies (GLS) by 144 learners who knew between two and twelve languages.Similarly to Miβler (2000), she revealed that the more languages the participants knew, the more likely they were to fall back on strategies for learning grammar, suggesting the existence of what she describes as a "threshold effect," whereby the use of GLS is proportional to the number of additional languages that are used by the learner.These findings were, by and large, corroborated in several subsequent empirical investigations.Psaltou-Joycey and Kantaridou (2009), for example, also used the SILL and found in the case of 1555 universitylevel Greek students learning foreign languages that learners who were trilingual used more LLS than bilingual ones, with those more advanced reporting using metacognitive and cognitive strategies more often.Sung (2011), in turn, detected a positive correlation between the number of languages learners knew and the frequency of LLS use, with those who had studied two or more L2s before starting to learn yet another additional language being more frequent users of metacognitive, cognitive, affective and social strategies than participants who had previously learnt only one language.The research project conducted by Jessner, Megens and Graus (2016) showed with the help of think-aloud protocols that, when exposed to a text in an unknown language (i.e., Romanian), young adult students who knew more additional languages were more likely to employ a greater number of compensatory strategies and were more creative when it came to tackling communication problems.
While the studies mentioned above are insightful in that they demonstrate that experience in learning multiple foreign languages translated into greater LLS use, at least with respect to some categories thereof, of particular relevance to the present paper are empirical investigations that have focused more directly on the application of strategies in specific languages learnt as an L2 or L3, also looking into learner-internal and learner-external factors impacting the learning process.In one such study, Merkelbach (2011) administered the SILL to investigate the use of LLS among Korean university students learning English as an L2 and German as an L3.The analysis revealed that L3 learners employed more metacognitive, memory, affective and social strategies, but this did not apply to compensation and cognitive strategies that were used more often by L2 learners.Importantly, the multilingual participants were highly motivated and deliberately selected to take part in the study, and previous experience in learning English as an L2 proved to be less important than having more than one first language (L1), a result that was attributed to a lack of strategies-based instruction in the English language classroom and thus little likelihood of transfer of LLS between languages.In the Norwegian context, Haukås (2015) conducted what is perhaps the first study of LLS use in L2 and L3 in a school setting where learners are required to learn specific foreign languages rather than allowed to choose them.She administered a slightly modified Norwegian version of the SILL to 132 learners of L2 English and 104 learners of L3 German and found, in contrast to the studies described earlier in this section, that the former reported using more strategies than the latter.She ascribes such results to limited awareness of the participants of how the knowledge of an L2 can benefit the learning of an L3, lack of motivation to learn German as well as limited utility of this target language (TL).Also worth mentioning are two studies undertaken in the Greek context, with Greek learned as an L2 and English as an L3, both of which used Greek versions of Oxford's (1990) SILL.In the first of them, Mitits and Gavriilidou (2016) examined 307 learners, aged 12 to 15, and found a correlation in the use of LLS in L2 and L3 but otherwise their results were inconclusive.While they demonstrated higher overall use of strategies in Greek than in English, the analysis of specific categories revealed more frequent application of cognitive strategies in L2 and affective strategies in L3.In an extension of this study, Mitits (2016) also compared the use of strategies by monolingual (L1 Greek) and multilingual learners (L2 Greek and L3 English), and was only able to detect an advantage for L3 learners in the case of compensation and memory strategies, but these results are of little relevance to the research project reported below.
When examining such research, one can hardly avoid the impression that the bulk of it has been conducted and interpreted with the aim of emphasizing the benefits of multilingualism with respect to the application of LLS and growing awareness of similarities and differences between different languages which can be capitalized upon in learning further languages.In fact, this goal of research on LLS within the framework of multilingualism is quite explicitly stated by Jessner and Török (2017), who write: "Current work on strategies makes clear that in terms of multilingual learning we have very limited knowledge of the actual processes and their nature.There is no doubt that work on multilingualism can help disentangle the interweaving and interdependence of metalinguistic and crosslinguistic awareness in processes of crosslinguistic interaction and offer food for further thought" (p.206).However, while there is undoubtedly merit to research projects focusing upon the use of strategies by multilinguals, it is not clear how merely demonstrating that someone who has experience in learning more than one additional language uses more strategies, both in general and with respect to specific categories, is likely to shed light on how metalinguistic awareness can in fact aid the learning process.In other words, although such research does lend support to the benefits of learning multiple languages, it is difficult to see how it could assist the learning of specific additional languages.For this to happen, more studies should be undertaken which would take account of the structural features of the languages learnt as L2, L3, L4, etc., or, like Merkelbach (2011) or Haukås (2015) did, consider the status and utility of these languages within a particular society or the role of individual factors such as motivation.After all, it could easily be argued that the frequency of LLS use as well as types of strategies may hinge upon the nature of a given TL (e.g., the size of the lexicon and the extent to which word-formation is based on compounding) or the proficiency in that language.Additionally, some languages might be inherently more appealing than others because of the way they sound, historical reasons or personal backgrounds, and the application of LLS can surely be a function of individual difference factors such as learning styles or the vision of one's ideal language self (see Dörnyei, 2009) with respect to a specific additional language.All of this indicates that research on the use of strategies in different languages has to go beyond merely providing evidence for greater frequency of LLS by multilinguals and should seek to determine the degree to which such use is also a function of the additional languages being taught and learnt.The study reported in the present paper represents a modest attempt to fill this gap by examining the LLS used by English majors in Poland who are required to achieve a high level of mastery of yet another additional language.

Aims and research questions
The study aimed to investigate the use of LLS in L2 and L3 by Polish university students enrolled in the first-year of a three-year BA program in English which also included an intensive course in an additional foreign language.The following research questions (RQs) were addressed: 1. Are there differences in LLS use in L2 and L3, both in general and with respect to the six categories of strategies included in the SILL and specific strategic devices?2. What are the dominant patterns in strategy use in L2 and L3? 3. Are there differences in strategy use between the most frequently learnt L3s?

Participants
The participants were 107 first-year students attending a regular BA program in English at a major Polish university, 86 of whom were females and 21 males.With the exception of one person who was an Erasmus student, all the participants were of Polish origin and spoke Polish as their mother tongue.Their mean age amounted to 20.3 years, and although it ranged from 17 to 24, the sample was rather homogenous in this respect, with the value of standard deviation equaling 1.72.For all of them, English, the language in which they were majoring, was an L2, with the mean experience in learning it standing at 12.34 and ranging from 5 to 17 years (SD = 2.72).When it comes to their mastery of English, it could be described as falling somewhere in between B2 and C1 according to the Common European Framework of Reference (CEFR), but there was much individual variation in this respect, particularly when different TL skills and subsystems were considered.The L3s the participants were also required to learn included: Spanish (41), German (38), French (14), Italian (7), Russian (5), Dutch (1) and Polish (1).The mean length of instruction in these languages stood at 4.61 years, but considerable heterogeneity could be observed, as it ranged from 1 to 14 years and standard deviation was very high (SD = 4.04).Not surprisingly perhaps, the proficiency in the L3s was lower than in the L2 and oscillated around A1 and A2 according to the CEFR, with some exceptions (e.g., there were participants who self-evaluated their mastery of Italian and Spanish as C1 and C2, respectively).It should also be mentioned that 52 (48.6%) students did not know any other languages in addition to the L2 or L3, while 55 (51.45%) participants knew one or more other languages, such as Spanish, German, French, Russian, Ukrainian, Chinese, Japanese or Latin.It can reasonably be assumed that they must have constituted their L3, L4 and so on but no detailed information is available in this respect.The BA program that all participants attended included an intensive course in English, classes in history, literature and linguistics, all of which were taught in that language, as well as a component devoted to L3 instruction which, however, was much less pronounced than in the case of the L2.

Data collection and analysis
Both quantitative and qualitative data were collected to provide insights into the use of LLS in L2 and L3.The former were obtained by means of the SILL (Oxford, 1990), which consists of 50 5-point Likert scale statements (1 -never or almost never true of me and 5 -always or almost always true of me) tapping the reported use of metacognitive, cognitive, memory, social, affective and compensation strategies.Although numerous criticisms have been leveled at this instrument (see e.g., Dörnyei, 2005;Dörnyei & Ryan, 2015;White, Schramm, & Chamot, 2007), including those by Oxford (2011Oxford ( , 2017)), who was the one to create it in the first place, its use appeared to be justified in view of the fact that it has been to some extent a default data collection instrument in the majority of studies of LLS use in L2 and L3 conducted to date.Thus its employment in the present investigation made it possible to attempt comparisons with the findings of previous research.In addition, as stressed by Amerstorfer (this special issue), despite its undeniable limitations, the SILL is definitely not past its expiration date and can still offer useful insights into the use of strategies, particularly when it is augmented with other research methods, as was the case in the present study.
Given the differences in the level of mastery of English among the participants and to reduce the danger of misunderstanding of some of the items included in the SILL, the inventory was translated into Polish (with the exception of one student for whom Polish was not the L1) and the internal consistency reliability of this translated version was satisfactory, both for the L2 and L3 (α = 0.81 for L2 and α = 0.73 for L3).The participants were requested to fill out the SILL during one of their classes, indicating the frequency of LLS use for English and the L3s that they were studying (different for different participants).Quantitative analysis was used in this case, with the means and standard deviations 1 being calculated for the L2 and the L3 for individual items, the six categories and the entire inventory.Oxford's (1990) guidelines were followed for interpreting frequency of LLS use as high (5.0-3.5),medium (3.4-2.5) or low (2.4-1.0).Two-tailed paired-and independent-samples t-tests were calculated to establish statistical significance of LLS use between L2 and L3, and between the most frequent L3s, respectively.
The quantitative data were complemented through semi-structured interviews in which four students learning Spanish as the L3, all of whom were volunteers, were queried about similarities and differences in the employment of strategies in L2 and L3, taking as a point of reference the items included in the SILL.As was the case with the entire sample, all of those students were much more proficient in English (L2, B1-B2 according to the CEFR) than Spanish (L3, A1-A2 according to the CEFR), and, not surprisingly, had been learning the L2 much longer (8-12 years) than the L3 (1-4 years).Although the authors are fully aware that learners of the remaining L3s should have ideally been included in the interviews, this was not possible owing to difficulties in accessing them.The students responded to three broad questions about the differences in the ways they approached the learning of L2 and L3: (1) "Compare the way you usually learn a specific thing in L2 and L3 (e.g., vocabulary, grammar)," (2) "Can you see any differences between your L2 and L3 learning?,"(3), "Can you see any similarities between your L2 and L3 learning?"Depending on the situation, the three questions were augmented with some additional queries intended to prod the students to offer more details.The interviews were held in Polish to ensure that the participants would be able to express their ideas freely and precisely, they were conducted individually by one of the present authors, and they were audio-recorded.The recordings were subjected to qualitative analysis which focused on the actions and thoughts employed in the process of the learning of the specific foreign language.

Results
As can be seen from Table 1, the reported use of LLS for the L2 (English) proved to be higher than for the L3, with the differences reaching statistical significance.This applied as well to the overall use of LLS (3.45 vs. 3.01) and all the six categories included in the SILL, that is, in the order of the magnitude of the difference in means, memory (a difference of 0.88), metacognitive (a difference of 0.79), cognitive (a difference of 0. 74), social (a difference of 0.57), compensation (a difference of 0.19), and affective (a difference of 0.07) strategies.Applying the criteria suggested by Oxford (1990) and expounded above, the overall use of LLS in L2 and present a more nuanced explanation, saying that under specific conditions the use of means is acceptable for ordinal data.
L3 could be characterized as medium, which was the case for most categories.The exceptions were three groups of LLS in the case of L2, that is, cognitive (M = 3.76), social (M = 3.76), and metacognitive (M = 3.74), where the reported means exceeded the threshold of 3.5 and strategy use could thus be described as high.Another interesting difference between the employment of LLS in L2 and L3 is the fact that there was notably more individual variation in the case of the latter, both on the whole and with respect to all the six categories, with the values of SD being the highest for social (0.89) and metacognitive (0.84) strategies.
Table 1 The reported use of language learning strategies in L2 and L3 It is also revealing to examine the differences in LLS use between L2 and L3 with regard to specific strategies, particularly those that deviated from the overall pattern, that is situations in which their reported application was higher in L3 than L2.In the case of memory strategies, this was observed for making associations with sounds or images (L2 M = 2.87 vs. L3 M = 3.09) or imagining a situation in which a particular word could be used (L2 M = 3.38 vs. L3 M = 3.42), with both of these being significant (p < 0.05).With respect to cognitive strategies, differences in favor of the L3 could be detected in the case of repeating or rewriting new words several times (L2 M = 3.74 vs. L3 M = 3.81), dividing words into parts that could be understood (L2 M = 3.22 vs. L3 M = 3.39), and searching for cognates in L1 and the additional language (L2 M = 3.74 vs. L3 M = 3.81), but the difference reached significance only for the last item (t = -2.66,p < .01).When it comes to compensation strategies, more frequent use of LLS was reported for L3 in the case of guessing the meaning of unknown words (L2 M = 3.80 vs. L3 M = 3.87), the use of gestures (L2 M = 3.54 vs. L3 M = 3.83), and making up new words (L2 M = 2.18 vs. L3 M = 2.44), with the latter two differences being significant (t = -2.44,p < .01,and t = -2.02and p < .05,respectively).As to affective strategies, the use of LLS was higher in L3 in the case of the strategy of trying to relax when being afraid to use the TL (L2 M = 3.19 vs. L3 M = 3.24), noting nervousness in the process of language learning (L2 M = 3.47 vs. L3 M = 3.50), and writing about feelings in relation to language learning in a diary (L2 M = 1.22 vs. L3 M = 1.19), but none of these differences proved to be statistically significant.In the case of social strategies, a statistically significant difference was only revealed for asking the speaker to repeat or slow down when a misunderstanding arises (L2 M = 4.14 vs. L3 M = 4.33, t = 2.05, p < .05).There were no situations in which the reported use of any metacognitive strategy was higher in L3 than L2.Generally speaking, it seems that the LLS used more frequently in the L3 than in the L2 were simply more geared to the challenges that lower proficiency learners were likely to be faced with when learning an additional language.
Although the investigation of the differences in strategy use in different L3s was difficult due to the fact that there were considerable discrepancies in the numbers of students learning those L3s, the researchers decided to undertake comparisons in the case of the most popular of these languages, that is, Spanish, German and French (41,38 and 14 participants,respectively).The overall reported use of LLS for the three languages equaled 3.41 (Spanish), 3.51 (German) and 3.48 (French), with the differences being too minute to reach statistical significance.Predictably, the two-tailed independent-samples t-tests also failed to reveal statistically significant differences with regard to any of the six categories included in the SILL.
As regards the interviews, the quality of the data they yielded left much to be desired since the participants seemed to not only exhibit scant knowledge concerning strategies and how they can be beneficially used, but also showed little awareness of what the process of learning L2 and L3 involved and found it relatively difficult to describe what they actually did when learning either of the two languages.This situation was quite surprising because, as English majors, the students had attended courses in linguistics and language teaching methodology, and it necessitated posing some additional queries to encourage the interviewees to provide more comments and details.Nonetheless, qualitative analysis of the recordings allowed the researchers to make three crucial observations.First, all the students agreed that they were more likely to use LLS in the L2 than the L3, but they could not really explain why this was the case.Second, the students did not see many differences in the ways in which they learned L2 and L3, and when such differences were mentioned, they were typically attributed to greater proficiency in English (L2), a finding which should hardly come as much of a surprise.After all, while watching movies or TV series with the original soundtrack may appear commonplace and natural for a student representing the B2 or C1 level, it surely must pose a major challenge for someone at an A1 level and may not even be attempted.On the other hand, the participants made no reference to other factors, such as the nature of the TL, its status or motivation to actually learn it.Third, irrespective of whether L2 or L3 was in focus, the scope of the reported LLS was very limited and the participants predominantly mentioned drawing on quite traditional memory and cognitive strategies, such as memorizing word and rules, and engaging in formal practice.There was little evidence for the application of metacognitive, social, affective or compensation strategies in the interview data.The excerpts that follow illustrate some of these points: I do not know what my language learning depends on.

I do not plan my learning.
I learn English by heart and in Spanish it is the same.
In English I watch many films and TV series, everything in English.In Spanish -not yet.
I learn Spanish grammar in Polish and I write some sentences in order to acquire it faster.In English I read books and do exercises because the grammar is more advanced.I learn differently.In English it is enough to read something and I remember it.I need to devote more time to learning Spanish.
In English I write the words down and rewrite them in order to memorize the spelling, in Spanish I write the words down and read them.

Discussion
Based on the analyses presented in the previous section, an attempt can be made to offer at least tentative answers to the research questions posed for the present study.With respect to RQ1, it was found that the use of LLS, both in general and with respect to the six categories included in the SILL, was higher in English (L2) than in the L3s that the participants were also learning.This by and large stands in contrast to the findings reported by Merkelbach (2011) as well as Mitits and Gavriilidou (2016), but mirrors the results obtained by Haukås (2015).Even though the students reported statistically significantly more frequent use of some strategic devices in L3 than L2 (i.e., creating associations with image or sound, envisioning a situation in which a particular word could be used, looking for cognates in the L1, relying on gesticulation, resorting to word coinage, asking interlocutors to repeat or slow down when a communication problem arises), these differences can primarily be ascribed to the overall lower proficiency in the L3.This is because, when using an additional language, a beginner learner is simply much more likely to try to find words in the L1 that are similar to those in the TL or use gestures to get the intended messages across in the face of inadequate TL resources.These findings were largely corroborated in the interviews because, even though the students mainly focused on similarities in learning L2 and L3, the differences they mentioned were clearly related to the proficiency level.What should also be noted is that there was more individual variation in LLS use in L3 than L2, which, yet again, is not entirely unexpected since the participants had been learning their L2 (English) for a much longer period of time and they were in fact majoring in it, which may have resulted in more consistent use of language learning strategies.There are a few viable explanations for such findings.For one thing, there were probably considerable differences in the participants' motivation to learn the L2 and the L3 (see also Henry, 2011), which, as demonstrated by Merkelbach (2011) and Haukås (2015), may be a crucial factor in the readiness to employ LLS.After all, it is clear that students majoring in a given L2 are likely to be much more motivated to improve their proficiency in that language rather than in an L3 that is often imposed on them by the requirements of the program.Second, and closely related to the previous point, there is the issue of the perceived utility of the TL being learnt, because, having chosen English as their major, the participants must have been cognizant of the benefits of their proficiency in that language for their future professional careers.By contrast, the benefits accruing form the command of the L3 may have been seen as much less tangible and thus less likely to become a stable element of the students' ideal selves (Dörnyei, 2009).Third, the overriding factor accounting for differences in the application of strategies, whether in regard to quantity or quality, was the substantial gap in L2 and L3 proficiency.On the one hand, differential proficiency levels may have explained the greater overall frequency of strategy use in the L2 but, on the other hand, it may have also been the reason why the use of some strategic devices was reported more frequently for L3, being more suitable to learning and using an additional language that is still relatively little known.Obviously, the nature of a particular L3 is also a crucial variable but it was difficult to explore in the present study because the L3s were not investigated separately and the data collection tools may not have been appropriate to capture the role of their distinctiveness with regard to LLS use.
When it comes to RQ2, few differences were revealed in the patterns of LLS use in the L2 and L3 since in both cases the students predominantly drew on a repertoire of quite traditional strategies, mainly memory and cognitive in nature, such as memorization or repetition, giving little attention to metacognitive, affective, social or cognitive strategies.This can perhaps best be explained in terms of instruction the participants received and evaluation procedures they underwent.This is because even though the courses in L2 and L3 are aimed to develop a high level of communicative ability, considerable weight is given to the mastery of pronunciation features, the development of a rich lexicon, and the command of a range of some-times very complex grammatical structures, with the instructional techniques employed being often very traditional and relying on translation, sentence completion or paraphrasing.Additionally, the regular in-class tests and the end-of-the-year examinations that the students are required to take usually stress form-focused components which to a large extent determine the final grades or scores.In such a situation, as Pawlak (2012) indicated with respect to grammar learning strategies, an intimate correspondence is bound to occur between how students learn and how they are taught and tested.As already pointed out above, when differences in patterns of LLS use were visible, they were mainly related to proficiency whereas other factors, such as those related to the nature and structure of a particular TL, seemed to take the back seat.What should also be emphasized is the scant awareness on the part of the students of the process of L2 and L3 learning, which translated into difficulty in obtaining requisite data.This situation also testifies to the fact that the program the students had been attending may have failed to reach its envisaged objectives and developed in the students the necessary level of understanding of both language and language learning.Finally, with respect to RQ3, no differences in strategy use were detected between the different L3s, which is surprising because the existence of differences of this kind could be expected on various grounds, related for example, to the utility of a given additional language, its perceptions by students, as well as the wider social or historical considerations (e.g., in the case of German).Truth be told, however, the data collection tools used may have not been sensitive enough to allow the researches to gain insight into the impact of such potentially important factors.
Although the study has produced invaluable insights into the use of LLS in L2 and L3, thus contributing to the scant body of empirical evidence in this respect, it also suffers from a number of limitations that dictate that the findings should be taken with considerable circumspection.First, although the SILL has been the obvious choice in the studies dealing with the application of strategies in L2 and other languages learners might be familiar with, the instrument suffers from a number of shortcomings, not least those related to the extremely general character of the items it comprises (cf.Amerstorfer, this volume; Dörnyei, 2015, Dörnyei & Ryan, 2015;Oxford, 2011;Tseng, Dörnyei, & Schmitt, 2006;Woodrow, 2005 ).Second, mainly owing to difficulty in accessing the potential informants, there were major discrepancies in the numbers of students learning different L3s, with the effect that these languages were treated as a single entity which they surely were not.Thus, it was not possible to look into LLS use as a function of a particular foreign language, whether with respect to its difficulty, structure, status in the Polish society or perceived utility.Third, again due to logistical constraints, it was not possible to include in the interviews representatives of all the L3s investigated with the help of the SILL, which clearly restricts the validity of the findings.Fourth, the study did not tap into the use of LLS with respect to concrete activities, whether from-focused (e.g., translation of sentences) or meaning-focused (e.g., finding differences between two pictures showing a scene in a park), which made it difficult, if not impossible, to capture the impact of the specificity of a particular TL on strategy use.Finally, the use of LLS can be a function of students' pursuing their personal agendas or possessing distinct individual profiles, but individual variation of this kind was not explored in the present study.

Conclusions and implications for future research
The present paper has reported a study intended to tap the use of language learning strategies in L2 and L3 among English majors with the help of the SILL and interviews with selected participants.Although the research project is not free from serious limitations, the analysis showed that the participants were more likely to use LLS in the L2 than L3, both on the whole and with respect to specific categories, a result that was mainly attributed to the impact of motivation, as English, the L2, was the TL in which the participants were majoring.In the case of the few strategies where the opposite was the case, the impact of proficiency was evident, with the LLS more often applied in the L3 being more suitable for less advanced learners.Other than that, few differences in patterns of strategy use in the two languages were uncovered, with the LLS being on the whole rather traditional, which may be the corollary of the instruction that the students received and the format of final examinations.Quite surprisingly though, as transpired from the interviews, the participants manifested little awareness of what learning additional languages involved and found it exceedingly difficult to pinpoint differences in strategy use in the L2 and L3.While these findings are promising and, in line with Merkelbach (2011) and Haukås (2015), help advance the research agenda beyond merely showing that multilingual learners are more strategic by focusing on the impact of various TLs in this respect, the results reveal just several pieces of the puzzle.There is clearly an urgent need for more research.Empirical investigations of this kind should focus more specifically on the employment of LLS in different L3s, target other populations than students majoring in a particular foreign language, explore LLS in L2 and L3 with respect to different skills and subsystems, compare the application of strategic devices in different kinds of language tasks (e.g., translation, or focused communication tasks that necessitate the application of a particular TL form for successful completion), and take into account the mediating influence of individual learner differences.On a somewhat different tack, the findings seem to indicate that even students majoring in English or other foreign languages should be made more aware of what the task of language learning involves and how the process could be enhanced with adept use of language learning strategies.Obviously, basing pedagogic intervention of this kind on the tangible research results on LLS use in L2 and L3 would be a no-lose proposition, which only stresses the pressing need for well-designed research projects in this area.