Investigating Learner Preferences in the Application of Language Learning Strategies: A Comparison between Two Studies

The general topic of this article is language learning strategies, i.e. activities that foreign language learners purposefully choose to improve their skills in a foreign language and to reach desired learning outcomes. To highlight the importance of combining quantitative and qualitative information in a study about strategic language learning, the article uncovers some weaknesses that may be created by a one-sided perception. It hence looks exclusively at the quantitative data gained in a mixed-methods study (Amerstorfer 2016) that used a popular strategy inventory (SILL; Oxford 1990) to collect numerical information about the language learning strategies of five English as a foreign language (EFL) learners. The core of the article is a comparison of the quantitative data with three strategy categories that resulted from a previous study (Griffiths 2013). The article closes with the statement that a one-sided perception is not appropriate for a study about strategic foreign language learning.


Introduction
Language teachers, teacher educators, and researchers have inquired into language learning strategies for forty years.Numerous projects employing quantitative, qualitative, and mixed research methods have produced countless publications addressing terminological definitions, as well as theoretical aspects like taxonomies, and practical applications like strategy-based instruction.This paper is an extract of a doctoral dissertation that studies language learning strategies in the context of cooperative learning.It will demonstrate the self-perception of strategic behaviour of five foreign language learners during cooperative English lessons.The article is based on a case study conducted at an Austrian secondary school.It aims to assess the similarities and differences of each participant's strategy use in comparison with three strategy types identified as significant for certain achievement levels (Griffiths 2003(Griffiths , 2013)).The comparison will reveal ambiguous results.It will further highlight the importance of clearly defined terminology and of carefully selected research tools and methods.
First, past research into language learning strategies will be briefly summarised.Second, the research design, environment, and methodology will be described with reference to the purpose of this article.Third, the data collected will be analysed and compared, which will finally lead to some reflections highlighting the importance of acknowledging individual learner differences.

Research into language learning strategies
Ever since Rubin's first mention of language learning strategies (LLSs) in 1975, the topic has been of immense interest to scholars around the globe.In the past forty years heated discussions about appropriate terminology, categorisation schemes, and practical applications of LLSs have occupied the academic community (for example , Hosenfeld 1976;Naiman et al. 1978;O'Malley et al. 1985;Wenden 1987;O'Malley and Chamot 1990;Oxford 1990Oxford , 1999Oxford , 2011;;Weinstein et al. 2000;Griffiths 2008Griffiths , 2013;;Cohen 2012).Consensus has been met regarding the general notion of LLSs by describing them as actions undertaken by learners to support learning processes and/or to produce language output.
A milestone in the development of LLS research is Oxford's (1990) categorisation system in which she divides LLSs into direct strategies (memory, cognitive, and compensation strategies) and indirect strategies (metacognitive, affective, and social strategies).The general distinction between direct and indirect strategies was first drawn by Rubin (1975Rubin ( , 1981) )  Inventory for Language Learning (SILL), which is one of the most commonly used instruments in the field of LLS research to date.The SILL consists of fifty statements about LLS usage that are rated by participants on a 5-point scale from 1, never/almost never true of me, to 5, always/almost always true of me (see table 1).The statement ratings enable the calculation of participants' average overall LLS use as well as an average application of strategies belonging to the six strategy types (A = memory, B = cognitive, C = compensation, D = metacognitive, E = affective, and F = social strategies).
The overall aim of the SILL is to identify and quantify LLSs based on learners' selfperception.
Oxford nominated high, medium, and low frequency of strategy use as follows:

High
Always or almost always used 4.5 to 5.0 Usually used 3.5 to 4.4 Medium Sometimes used 2.5 to 3.4

Low
Generally not used 1.5 to 2.4 Never or almost never used 1.0 to 1.4 Table 1: Key to determine high, medium, and low frequency of LLS use by Oxford (1990) The SILL profiles below quantitatively demonstrate how the participants of the study described in this paper ranked different strategy types (parts A-F) in sum and on average.
Oxford's key corresponds with average ratings for each strategy type.An overall sum of all ratings and an average rating of all 50 statements are given at the bottom of each participants' SILL profile.
In her PhD thesis, Griffiths (2003) used the SILL to investigate whether there is a relationship between highly frequently used LLSs and successful language learning.Griffiths developed SILL profiles of 348 participants aged 14 to 64 from 21 different nations.
The group under investigation consisted of approximately one-third male and two-thirds female participants with proficiency levels ranging from elementary to advanced (Griffiths 2013).Griffiths focused on strategies most frequently used by elementary learners and by advanced learners, as well as on strategies applied highly frequently across all proficiency levels.In comparison to the elementary language learners, the advanced participants reported an overall higher frequency in LLS application.Additionally, the study revealed that advanced learners reported using nine times as many different LLSs than elementary learners.Based on these findings, Griffiths identified three groups of LLSs (see This article presents the quantitative results of a recent case study (Amerstorfer 2016) in comparison with Griffiths' (2013) findings about base, core, and plus strategies.
Specifically, it focuses on base and plus strategies because core strategies, according to Griffiths, are used by learners of all proficiency levels.Before the comparison will be made, the research design, environment and methodology of the case study will be explained.
3 Research design, environment, and methodology The case study at hand investigates five learners' strategic behaviour during cooperative EFL lessons.All participants are female and between 14 and 18 years of age.To protect anonymity, pseudonyms are used.The participants' shared first language (L1) is German.In this paper, a clear distinction is drawn between proficiency levels and achievement levels.Proficiency levels are regularly assessed in school and defined by the CEFR 2 .The participants' achievement levels, on the other hand, reflect the grades they received in the subject English in the semester reports (table 3).The high-level achievers in this study received A-grades, while the low-level achievers received D-grades.In other words, the achievement level describes how well the pupils performed in English in the first semester of the academic year 2013/14.The research environment of the study is CoOperative Open Learning (COOL), a statefunded teaching concept that was introduced in 1996 in Austri (Cooltrainers, n.d.).The teaching philosophy underlying COOL is the Dalton Plan 3 (Parkhurst 1922), a humanistic teaching approach developed in the early 1920s.In COOL schools, pupils take increased responsibility for their learning processes and outcomes.About a third to half The starting point of this paper is the five participants' self-perception regarding their use of language learning strategies, which will be demonstrated in their SILL profiles (tables 4, 6, 8, 10, and 12).Second, the strategies that were rated highest and lowest will be isolated to enable a comparison with Griffiths' ( 2013) findings (tables 5, 7, 9, 11, and 13).The purpose of this paper is, therefore, to analyse how each participant's self-perception of LLS use compares to Griffiths' categories.

Analysis of data
For a comparison with Griffiths' (2013) study, this section will summarise and analyse the quantitative results of each participant.Only the highest and lowest SILL ratings are taken into account.An analysis of the results reveals a few problems.First of all, the originally designed SILL (Oxford 1990) is outdated.Some strategies that are often used by pupils in schools today are not included in the original version of the SILL, particularly those related to modern technologies.Second, participants' responses to the SILL are purely based on self-perception and memory.Research tools that solely depend on a person's own judgement and recollection lack reliability and may not reflect reality.Third, the SILL provides only quantitative results.While this may be unproblematic in a large-scale study, it is inappropriate for a case study with a small sample size.The three issues raised call for an inclusion of other research methods.

Christina (low-level achiever)
Through semi-structured initial interviews and recorded lesson observations combined with stimulated recall interviews, the shortcomings mentioned above can be overcome (see Amerstorfer 2017 for more information).Strategies that are not included in the SILL can be added, and numerical results supplemented with qualitative information.The purpose of this article, however, is to compare the quantitative findings generated through the SILL with Griffiths' categorisation of base, core, and plus strategies.Therefore, all qualitative aspects have been neglected.As was mentioned in the introduction, this article is a snapshot of a larger study.The additional research methods mentioned in this paragraph are, in fact, applied in the study (Amerstorfer 2016) but not reported here.
The data analysis notably reveals that Griffiths' study exhibits some weaknesses.First, there is a terminological issue that needs clarification.On the one hand, Griffiths inquires about the relationship among successful language learning and certain language learning strategies and strategy types.On the other hand, learner proficiency levels are mentioned in the analysis of the results.As was explained above, proficiency levels should clearly be distinguished from achievement levels.To clarify, while two learners can be on the same proficiency level (for instance, B1 according to the CEFR), one can perform well in the school subject English and receive an A-grade in the semester report, while the other one can be a low-level achiever and get a D-grade because overall they did not do so well in that particular semester.Such terminological ambiguities can lead to misinterpretations of results.Second, only highly frequently used strategies are analysed in Griffiths' study.Strategies with low ratings are disregarded.This raises a number of questions, for instance, what can low-rated SILL statements say about a learner's language learning strategy use?Are certain strategies avoided by learners?If yes, why?Third, according to Griffiths, core strategies are used by learners across all levels, which again poses questions about the meaning of low or high ratings for the corresponding SILL statements in the core category.In a SILL-only approach, a comparison between high-level and low-level achievers' application of core strategies becomes irrelevant.However, adding a qualitative component can reveal interesting results, such as reasons for certain strategy choices or connections between single strategies (Amerstorfer 2016).Fourth, how should SILL statements that are not included in any of Griffiths' categories be handled?In order to draw a valid comparison, they would have to be excluded from the study.That would, however, cause additional complications because high or low ratings for them also convey meaning.
All of these factors lead to a high degree of ambiguity in the comparison between Griffiths' study and the SILL results extracted from the case study reported in this article.As was mentioned before, the study mixes a number of research methods in order to prevent flaws resulting from a SILL-only approach.A combination of SILL outcomes with qualitative information leads to comprehensive findings that acknowledge individual learner differences in the application of language learning strategies.Nevertheless, the purpose of this article was a comparison between the participants' SILL results and Griffiths' categories (2013).
and later expanded by Oxford.Figuratively speaking, Rubin created the skeleton and Oxford put flesh on the bones.Oxford's comprehensive classification demonstrates what each of the six sub-categories encompasses and provides a total of 62 examples.Based on this categorisation scheme, Oxford developed the Strategy of a COOL timetable consists of open learning periods (COOL lessons) in which pupils work independently of a teacher.The remainder are regular lessons that are teachercentred, similar to those in mainstream schools in Austria.The class teachers involved in COOL meet regularly to discuss any current business and to design assignments and tasks for COOL lessons.During those lessons, pupils are encouraged to work collaboratively, support each other, and offer feedback on each other's work.They self-regulate their learning and take responsibility for the learning progress.Pupils in COOL schools are aware that what they cannot finish in school must be completed in their leisure time.

Table 3 :
Participants' proficiency and achievement levels

Table 6 :
Sabrina's SILL profile Sabrina's SILL profile shows that her overall strategy use is mediocre (2.7).She prefers metacognitive and affective strategies (both 3) and uses memory strategies least fre-Out of the five participants, Stella shows the most similarity to Griffiths' results.She ranked two strategies (items 6 and 43) lowest which both belong to the base category and are typically used frequently by low-level achievers.Seven of the nine strategies with the highest ranking are plus strategies(items 1, 10, 19, 22, 27, 39, and 40), which is indicative of high-level achievers in Griffiths' study.The two remaining strategies ranked highest belong to the base category (item 9) and to the core category (item 29).All of the strategies ranked highest or lowest by Stella were mentioned in Griffiths' study.Accordingly, Stella represents the results of a typical high-level achiever.
disagreement Table 5: Christina's lowest-and highest-rated SILL statements 5 in comparison withGriffiths (2013)In contrast to Griffiths' view that base strategies are typically used by low-level achievers, the two strategies rated lowest by Christina (items 3 and 43) both belong to the category of base strategies.However, Christina's highest ratings for two other base strategies (items 8 and 34) are in line with Griffiths' results.Christina further rated three core strategies highest (items 25, 32, and 38).Core strategies were found to be used by all pupils, so they are not relevant in this comparison.Christina's high rating for two plus strategies (items 10 and 40) deviates from Griffiths' findings because those are believed to be usually representative of high-level achievers.One strategy (item 35) rated highest by Christina is not mentioned in any of the categories.With two strong agreements (items 8 and 34) and four strong disagreements (items 3, 10, 40, and 43) Christina's strategy use does not represent that of Griffiths' typical low-level achiever.quently(2.222).According to Griffiths' study, low-level achievers favour memory strategies, but Sabrina is the opposite.Sabrina's lowest-rated strategies consist in equal numbers of base strategies, which Griffiths found typical of low-level achievers (items 6, 7, 26, and 43), plus strategies, which are typical of high-level achievers(items 14, 18, 20, and 48), and strategies that were not mentioned inGriffiths' scheme (items 5, 16, 35, and 49).No core strategies were rated lowest by Sabrina, but all six highest-rated strategies are core strategies, which are used by learners across all levels.A remarkably large number of high ratings, as in Sabrina's case, cannot be satisfyingly compared to Griffiths' study because core strategies were not found representative of either achievement group.The analysis of Sabrina's quantitative data does, therefore, not show any particularly significant similarity with Griffiths' results.The opposite is the case for the four statements rated lowest by Sabrina that were described as characteristic of low-level achievers.All in all, Sabrina's strategy use is not representative of a typical low-level achiever.

Table 14 :
Numerical agreements and disagreements with the strategy types defined byGriffiths (2013)