EFFECTS OF LEARNERS ’ LANGUAGE PROFICIENCY ON THEIR LANGUAGE LEARNING STRATEGIES

Language proficiency is one prominent factor that could affect the use of language learning strategies (LLSs) in learning English as a second language among the learners. The influence of language proficiency on the use of LLSs might differ as language learning has extended to the contexts of learning other than the learners’ differences. With that in view, this study seeks to examine various types of LLSs based on the language proficiency of 1699 learners in a Malaysia tertiary context. This study utilised the adapted version of Strategy Inventory for Language Learning (SILL, 1990) Findings revealed that high proficiency learners employed LLSs from high to medium level whereas average and low learners showed medium strategy use of all LLSs. One-way ANOVA tests further indicated that there were significant differences in the use of direct strategies based on their language proficiency. It was also found that language proficiency had a significant influence on the use of cognitive, compensation, social, metacognitive and affective strategies. Positive to mix variation of the high strategy use for individual items of SILL with significant differences were reported. Implications of this study were also discussed.


Introduction
In Malaysia, English is considered as a second language after the national language, Bahasa Melayu.Even though English is widely used in Malaysian education, average Malaysian students are still unable to master this language adequately in the aspects of verbal fluency, writing compositions and using proper grammar, which resulted in unsatisfactory examination results (Kiram, Sulaiman, Swanto, & Din, 2015).This has created problems for "academically able students" to enrol in universities because good command of English is regarded as an advantage for students to access information and all kinds of knowledge, which are mostly written in English (Kiram et al.,p. 2).Such issue on poor language proficiency should be of concern to especially the educators, scholars and policy makers.This is due to the reason that having good communication skills in English will ensure high employability rate among these undergraduates because they will be only able to perform effectively to meet the challenging nature of current workplace communication if they possess high proficiency in the English language (Moslehifar & Ibrahim, 2012).Similarly, Navaratnam (2014) also asserted that good proficiency in English enhanced graduates' worth for employment especially in the corporate world.Phang (2006) further stated that poor command of English language had caused the college graduates to remain unemployed (as cited in Kassim & Ali, 2010, p. 168) despite the numerous efforts by many relevant parties to increase the standard and language proficiency of the learners and to produce graduates who possess the ability to communicate and work effectively at the workplace.With these concerns, this paper aims to discuss the use of LLSs based on their language proficiency as exposure on how to use LLSs appropriately in various contexts would resulted in a more successful language learning process.

Literature review
There are various definitions to describe LLSs.For example, Oxford (1990) defined LLSs as the specific actions or techniques used by students consciously to improve their second language learning skills.It is also acknowledged that the classification of language learning strategies by Oxford (1990) is the most widely recognised and employed (Huang & Nisbet, 2014).In addition, Griffiths (2013) also agreed that the taxonomy of strategies by Oxford was comprehensive.Oxford (1990) classified these strategies into direct and indirect strategies.Direct strategies are the strategies that require mental processing of the language which consist of memory strategies, cognitive strategies and compensation strategies.Memory strategies, also known as mnemonics, are used to remember and retrieve new information.Memory strategies include four sets of strategies, like "creating mental linkages, applying images and sounds, reviewing well and employing actions" (Oxford, 1990, p. 38).Cognitive strategies are used to understand and produce the language.Cognitive strategies include four sets of strategies, namely "practicing, receiving and sending messages, analysing and reasoning, and creating structure for input and output" (Oxford, 1990, p. 43).Compensation strategies enable the learners to use the language irrespective of their knowledge gaps.Compensation strategies include two sets of strategies, namely "guessing intelligently in listening and reading" and "overcoming limitations in speaking and writing" (Oxford, 1990, p. 47).On the other hand, indirect strategies are strategies used to "support and manage language learning" without direct involvement with the target language (Oxford, 1990, p. 135).These strategies are further classified into metacognitive, affective and social strategies.Metacognitive strategies enable learners to "control their own cognition" through "centering, arranging, planning and evaluating to coordinate their learning process" (Oxford, 1990, p. 135).Metacognitive strategies include three strategy sets, namely "centering your learning, arranging and planning your learning, and evaluating your learning" (Oxford, 1990, p. 136).Affective strategies are used to regulate the emotions of the learners whereas social strategies encourage "learning with others" (Oxford, 1990, p. 15).Affective strategies consist of three sets, namely "lowering your anxiety, encouraging yourself and taking your emotional temperature" (Oxford, 1990, p. 140).As for social strategies, these strategies comprise three sets of strategies, namely "asking questions, cooperating with others and empathising with others" (Oxford, 1990, pp. 144-145).
Many factors were found to affect the choice of LLSs in learning English among learners, namely degree of awareness, stage of learning, task requirements, teacher expectation, age, sex, nationality/ethnicity, general learning styles, personality traits, level of motivation and the purpose of language learning (Oxford, 1990, p. 13).Despite that, language proficiency is among one of the prominent factors that could affect the use of LLSs.Past literature on the effects of language proficiency reported mix conclusions.Some results apparently agreed that language proficiency had influenced the use of LLSs statistically while some others offered contradicting findings that language proficiency failed to affect the use of LLSs statistically.According to Takeuchi, Griffiths and Coyle (2007), a few possible reasons could explain why language proficiency affected language learning used differently (as cited in Kamalizad & Samuel, 2015).These possible reasons could be due to "other variables that might overshadow language proficiency factors, the type of instruments used, the strategies used differed from the strategies in SILL and the flexibility in using the strategies in the specific contexts" (as cited in Kamalizad & Samuel, 2015).In addition, a few researchers (Green & Oxford, 1995;McIntyre, 1994;Rees-Miller, 1993) asserted that the association between LLSs and language proficiency should be taken with caution regardless of the common notion that learners with higher language proficiency applied strategies more effectively (as cited in Lai, 2009).For example, Rees-Miller pointed out that there was no particular direction in this causal relationship (as cited in Lai, 2009, p. 259).Other than that, Green and Oxford (1995) and McIntyre (1994) further suggested that strategies are both the cause and outcome of improved language proficiency (as cited in Lai, 2009, p. 259).
Learners with better language proficiency were found to employ LLSs more frequently than the less proficient ones (Alhaisoni, 2012;Gerami & Baighlou, 2011;Salahshour, Sharifi, & Salahshour 2013;Zhou & Intaraprasert, 2015).For example, Zhou and Intaraprasert's study (2015) indicated that the preservice teachers with high language proficiency highly employed LLSs significantly compared to those with lower proficiency levels.Likewise, the research by Gerami and Baighlou (2011) showed that successful EFL students used a wide range of strategies compared to unsuccessful ones.Successful students also preferred the use metacognitive strategies while the less successful ones used more of surface level of cognitive strategies.It was also found that successful students employed overall strategies significantly more frequently than the unsuccessful ones.S alahshour, Sharifi and Salahshour's study (2013) revealed that learners with higher language proficiency frequently employed all types of strategies by Oxford (1990) if compared to lower proficiency learners.Other than that, it was found that high proficiency group employed metacognitive and social strategies significantly even though significant differences were found in the use of overall language learning strategies between these two groups of learners.The study by Alhaisoni (2012) among Saudi EFL undergraduates revealed that higher proficiency learners employed all the six categories of strategies more significantly than lower proficiency students.
Past literatures which reported that language proficiency had statistically significant effect of the use of LLSs by learners could be seen in some studies (Al-Qahtani, 2013;Ghadirzadeh, Hashtroudi, & Shokri, 2013;Ghavamnia, Kassaian, & Dabaghi (2011); Goh & Kwah, 1997;Kiram et al. (2015); Kunasaraphan, 2015;Platsidou & Sipitanou, 2015;Zhou & Intaraprasert, 2015).For example, the study by Kunasaraphan (2015) revealed that there was a significant difference in the use of the overall LLSs and the five specific strategies, namely memory, cognitive, compensation, metacognitive and social strategies on the first year students with different levels of English proficiency.Another study by Zhou and Intaraprasert (2015) also revealed that the frequency of the use of overall LLSs differ significantly based on their language proficiency.Platsidou and Sipitanou's study (2015) among elementary and middle school students showed that there was significant correlation between all the LLSs and their perceived language proficiency and with the highest indices observed between the metacognitive and cognitive strategies and their perceived language proficiency scores.In addition, higher perceived language proficiency learners were positively and significantly related to their higher frequency use of LLSs (Platsidou & Sipitanou, 2015).Al-Qahtani's study (2013) among the first year students revealed that there was a significant relationship between the grade in English language proficiency and the social strategies.Other study by Kiram et al. (2015) revealed that significant effect was only found between cognitive strategies and language proficiency even though there was no significant relationship between the overall LLSs and language proficiency.Similarly, the research by Goh and Kwah (1997) also found significant differences in the use of cognitive and compensation strategies based on three levels of proficiency of students at a university in China.
A unique study by Kamalizad and Samuel (2015) revealed that the role of language proficiency on the use of LLSs by Iranian English learners showed different findings based on the contexts of learning, namely ESL setting (Malaysia) and EFL setting (Iran).Findings showed that there was significant difference on the learners for the overall use of SILL and the six strategy categories in the ESL setting.However, for the EFL setting, it showed no significant differences between the overall use of LLSs and the individual categories of LLSs.
Other studies also indicated that no significant differences were found in the use of LLSs based on language proficiency (Kiram et al., 2015;Namwong, 2012;Seifoori & Yaghchi, 2015;Tezcan & Deneme, 2016;Yilmaz, 2010).For example, no significant differences were found in the use of LLSs between the successful and average 8th grade young Turkish learners of English as a foreign language from three private primary schools (Tezcan & Deneme, 2015).Such non-significant difference might be due to other possible individual factors and social contexts of learners that resulted in fewer differences in the use of LLSs irrespective of their levels of English language proficiency.Another study by Seifoori and Yaghchi (2015) among EFL Iran students, majoring in teaching English at a university revealed that there was no significant difference in the use of LLSs and language proficiency.Seifoori and Yaghchi also mentioned that there was lack of literature on the possible differences between the high and low proficient learners.Other than that, they further asserted that there was a paucity of data on the relationship between LLSs and language proficiency.Similarly, Namwong's finding (2012) also stated that there were no differences in the LLSs used between high and low proficiency of fourth year students at a university.Such finding could be due the reason that these students had some background in learning English as they were majoring in teaching English to other learners.Likewise, the study by Yilmaz (2010) also found that no significant differences were found on the use of strategies based on their university's averages for English major students in Turkey.
There were studies (Bagheri, 2015;Ghadirzadeh, Hashtroudi, & Shokri, 2013;Griffiths, 2003;Soodmand Afshar, Tofighi, & Hamazavi, 2016) that indicated students who employed more of LLSs will achieve better language success.This view was in agreement with the claim by Griffiths and Oxford (2014), in which numerous studies had showed significant positive correlations in the use of LLSs and success in language learning.Griffiths ( 2003)'s study reported a statistical significant relationship between the rate of progress and the increase in the use of LLSs.This could be seen by Bagheri's study (2015) that revealed a positive relationship between the LLSs score and the language score for speaking and listening.In other words, this study had proven that students who employed more strategies would obtain better language result.Another study by Ghadirzadeh, Hashtroudi and Shokri (2013) also found that students highly preferred metacognitive strategies, which significantly affected their success in language learning.In a similar vein, the study by Soodmand Afshar, Tofighi and Hamazavi (2016) among EFL learners from English Language Institute revealed that there was a statistical significant positive relationship between the use of LLSs and the L2 achievement of these learners.
On the contrary, the use of LLSs did not necessarily have much impact of the language proficiency of learners based on the findings of some studies (Chand, 2014;Lai, 2009).For example, a study by Chand (2014) revealed that there was a positive weak relationship between LLSs and the academic language proficiency among the undergraduates.It showed that the strategy used did not have much impact of their proficiency in academic language.Furthermore, Lai (2009) claimed that the exact nature of the strategy cause and effect was controversial despite various studies that indicated the more proficient learners had better use of LLSs.In addition, there was no consensus yet on whether certain "strategy patterns" or "specific strategy types" were associated with effective learners.Other than that, according to Salahshour et al. (2013), language proficiency factor has not gained due attention even though it is generally regarded as one of the factors that could affect the use of LLSs.

Aims
This study aimed to investigate the LLSs employed by the learners in a public university in Malaysia based on the levels of language proficiency.It will address the following research questions: • What are the LLSs employed by learners based on their language proficiency?• Are there any significant differences in the use of LLSs based on their language proficiency?
• What are the patterns of variation of "high strategy use" category for individual item of LLSs employed based on their language proficiency?

Methodology
This study employed a quantitative approach to obtain the information on the use of LLSs by first year undergraduates in a public university, namely Universiti Malaysia Sarawak (Unimas) in Malaysia.

Subjects
There were 1708 first-year undergraduates from the eight faculties of Unimas participated in this study.These undergraduates comprised 85.14% of the population under study, excluding those 133 undergraduates, who were involved in the pilot study prior to this actual study.After dealing with nonresponse items through phone or e-mail, there were nine missing values of undergraduates' Malaysian University English Test (MUET) results.Hence, only 1699 (84.70%) undergraduates of the population were included in the subsequent statistical analyses.The MUET exam, which comprised six bands, was used to indicate the undergraduates' English language proficiency prior to their admission to Malaysian tertiary institutions (Ngah, Radzuan, Fauzi, & Abidin, 2011).As such, MUET band was used to indicate language proficiency of undergraduates in this study.There were 669 (39.4%) undergraduates who obtained MUET Band 1 or Band 2 were classified as low achievers whereas 951 (56.0%) undergraduates were average achievers with MUET Band 3 or Band 4. The remaining, 79 (4.6%) undergraduates who scored Band 5 or Band 6 was grouped as high achievers.

Instrument
This study utilised the adapted version of Strategy Inventory for Language Learning (SILL) ESL/EFL version 7.0, developed by Oxford in 1989 (Oxford, 1990, pp. 293-300) in order to elicit LLSs employed by first-year undergraduates.The underlying constructs for 50-item of self-reporting SILL survey questionnaire were memory, cognitive, compensation, metacognitive, affective and social strategies.
For each item, respondents were required to circle one of the 5-point Likert scale options, ranging from "never or almost never true of me" to "always or almost true of me".Mean scores were used to indicate the frequency of employing particular LLS by learners.Coefficient of Cronbach's alpha of overall 50item of SILL was high, .926.This coefficient was congruent with other past studies that showed high reliability of SILL across various cultural groups (Oxford, 1996).

Data analysis
Quantitative data from the survey questionnaire was analysed using SPSS (Statistical Package for the Social Sciences) software package.After obtaining descriptive statistics of mean score and standard deviation for each LLS, subsequent statistical techniques were employed for respective inferential statistics.The significance level, α was set at .05.One-way ANOVA (Analysis of Variance) was used to determine existence of statistical difference on the mean scores of each LLS based on levels of language proficiency.This was followed by post-hoc Tukey's Honestly Significant Difference (HSD) tests in order to identify possible significant differences among them.Chi-square test was then conducted with respect to particular SILL items.This non-parametric analysis was only carried out to compare the "high use" of SILL item based on the levels of English language proficiency of respondents which had significant differences on the use of LLSs.Based on 5-point Likert scale, SILL responses of 4 (usually true of me) and 5 (always or almost true of me) were categorised as "high strategy use" for valid analysis (Green & Oxford, 1995).This categorisation was done to meet the minimum number of respondents required in each cell of cross tabulation particularly for smaller sample size of respondents with high level of language proficiency.Larger sample with expected frequencies greater than five would have a better approximate chi-square distribution and accurate chisquare statistic (Field, 2009).Variation of significant SILL items would then be characterised as positive, negative or mixed based on the percentage of "high strategy use" between different levels of language proficiency of respondents (adapted from Green & Oxford, 1995).Positive variation was shown when there was an increasing pattern in percentage of "high strategy use" of a SILL item corresponding to better levels of language proficiency.Vice versa, corresponding decrease in percentage of "high strategy use" of a SILL item at higher levels of language proficiency indicated a negative variation.Neither of these two patterns would be termed as mixed.
Prior to all statistical analyses, outliers beyond lower or upper inner fence of k (interquartile range) with k = 1.5 (Hoaglin, Iglewicz & Tukey, 1986) of each language learning strategy were removed.

Results and Discussion
Table 1 shows the LLSs employed by the respondents based on their levels of language proficiency.It was found that all the LLSs were used in a moderate manner by both low and average achievers based on the classification by Oxford (1990) as "medium" users with scores ranging from 2.5 to 3.4.On the contrary, high achievers were "high" users for metacognitive and cognitive strategies (M ≥ 3.5) and "medium" users of the remaining strategies.The findings from this study were quite similar to the study by Madhumathi, Ramani and Prema (2014) among ESL below average students that used the overall language learning strategies at medium level.Their studies also showed that memory, affective, social, compensation and cognitive strategies were used in the medium level, except metacognitive strategy which was categorised under "low strategy use".
In this study, findings from independent samples t-tests showed that low achievers significantly employed more indirect (M = 3.23, SD = .65)than direct (M = 3.04, SD = .54)strategies in learning English (t(3852) = 10.137,p < .05).However, there were no significant differences found in the use of direct and indirect strategies by average and high achievers in learning English.Such findings contradicted the finding by Kunasaraphan (2015), which revealed that first year students employed more indirect strategies than direct strategies.Note.MEM = memory; COG = cognitive; COM = compensation; MET = metacognitive; AFF = affective; SOC = social; ">" = significant higher mean score; "/" = no significant mean scores at p < .05 Further statistical analyses through One-way ANOVA tests (Table 2) revealed that there were significant differences in the use of strategies in learning English based on respondents' levels of language proficiency, namely low achievers (F(5, 3966) = 42.390,p < .05),average achievers (F(5, 5637) = 96.702,p < .05)and high achievers (F (5, 456) = 20.455,p < .05).At α = .05,Tukey's HSD tests revealed low achievers preferred mostly metacognitive strategies, followed by social, affective and memory strategies in learning English.The use of cognitive and compensation strategies were significantly lower compared to metacognitive and social strategies, but no significant differences with affective and memory strategies.Average achievers also preferred most metacognitive strategies, followed by social/cognitive/compensation, memory and affective strategies.The use of social, cognitive and compensation learning strategies did not show any significant differences.For high achievers, they generally employed metacognitive, cognitive and social strategies interchangeably as their most preferred LLSs.The use of these three strategies was significantly higher than memory strategies and this was followed by affective strategies as the least preferred strategies.These respondents also significantly employed less compensation strategies compared to metacognitive strategies but more frequent than affective strategies.However, the use of compensation strategies had no significant differences with cognitive, social and memory learning strategies.Significant order of LLSs employed by respondents in learning English based on their levels of language proficiency could be summarised in Figure 1.Regardless of their levels of language proficiency, metacognitive strategies were significantly highly preferred by all respondents in learning English.Low achievers least preferred memory strategies but average and high achievers least preferred affective strategies in learning English.

Metacognitive
Thus, these findings also support the findings by Magogwe and Oliver (2007) whereby good and fair students mostly preferred the use of metacognitive strategies.This might be due to the reason that better students are considered to be more independent in their learning and are able to control and regulate their learning.Such features were relevant to the characteristics of metacognitive strategies.In addition, affective strategies that were the least preferred strategies among the average and high achievers was congruent with the finding by Goh and Kwah (1997).Less preference of affective strategies might be due to the culture of Asians whereby learners are viewed as shy and conservative learners.They are not encouraged to express their feelings in the learning process.Note.LLSs = language learning strategies; MEM = memory; COG = cognitive; COM = compensation; MET = metacognitive; AFF = affective; SOC = social; ">" = significant higher mean scores; "/" = no significant mean scores at p < .05 Based on Table 3, One-way ANOVA tests revealed that there were significant differences for direct strategies used by respondents based on their language proficiency.However, no significant differences were reported for indirect strategies based on their language proficiency.Tukey's HSD tests showed that both high and average achievers employed more direct strategies compared to low achievers in learning English.However, there was no significant difference between high and average achievers in the use of direct strategies.From the One-way ANOVA tests, five out of six LLSs employed by respondents in learning English showed significant differences, namely cognitive (F(2, 1652) = 51.172,p < .05),compensation (F(2,1696) = 21.010,p < .05),social (F(2,1690) = 6.256, p < .05),metacognitive (F(2, 1674) = 3.769, p < .05)and affective strategies (F(2,1673) = 30.467,p < .05).However, there was no significant difference in the use of memory strategies in learning English among respondents with different levels of language proficiency (F(2,1674) = 2.088, p > .05).With α = .05,Tukey's HSD tests showed that respondents with different levels of language proficiency had significant differences in the use of cognitive and affective strategies.Statistical findings revealed that when the respondents had better command of English language, they significantly preferred more cognitive and less affective strategies in learning English.Although there were no significant differences in the use of compensation and social strategies between high and average achievers, these learners significantly highly use both of these strategies compared to low achievers.High achievers also significantly utilised more metacognitive strategies compared to low achievers in learning English.
Finding also revealed that low achievers significantly did not employ certain language learning strategies as frequent compared to learners with better command of English.This could be reflected by lower preferences of low achievers in the use of cognitive, compensation and social strategies (Table 3).However, low achievers employed more of affective strategies significantly compared to other levels of achievers in pursuing English language.Vice versa, learners with high language proficiency had higher tendency to employ various types of LLSs in learning English, ranging from high to medium strategy used based on the classification by Oxford (1990).This could be indicated by significant higher mean scores of preferences for cognitive, compensation, social and metacognitive strategies by high achievers compared to low achievers.The finding of Nguyen and Godwyll (2010), in which learners with higher language proficiency had higher use of the cognitive, compensation, social and metacognitive strategies are consistent with the results in this study.This finding also further supports the results by other researchers (Embi, Long, & Hamzah, 2001;Griffiths, 2007;Oxford, 2002), in which high achievers appeared to use multiple strategies compared to low achievers.Based on statistical analyses of "high strategy use" (score 4 or 5) of SILL items by using chi-square tests (Table 4), 46% of the items (or 23/50) showed significant differences in the use of LLSs among respondents with different levels of English language proficiency (p < .05).Excluding memory strategies that had no significant difference of mean scores among respondents (Table 3), there were 11 SILL items related to cognitive strategies or 78.6% showed significant differences of "high strategy use" by the respondents (Table 4).This were then followed by SILL items for metacognitive strategies (55.6%), compensation strategies (50.0%) and affective strategies or social strategies respectively indicated similar percentage of 33.3%.From the 23 significant SILL items, 78.3% (n = 18) of SILL items showed positive variation while the remaining 21.7% (n = 5) SILL items showed mixed variation.None of the SILL item showed negative variation in this study.
90.9% of significant SILL cognitive items among respondents showed positive variation (Table 4).This result indicated that employment of those cognitive strategies increased with higher levels of English language proficiency.Respondents with better command of English tended to manipulate or transform their English learning.This was done by practicing their language learning through repeating strategies in speaking, writing (item no. 10, I10) and imitating the native speakers (I11).Besides, they also used recombining strategy in practising their English language (I13).Findings also revealed that more respondents with higher language proficiency skimmed the English passage first to gain quick ideas in receiving and sending messages (I18).They would avoid direct or verbatim translation in the aspect of analysing and reasoning the target language for their understanding or expression (I22).
Besides that, they were in favour of practising their English language in realistic settings to improve their listening skills (I15), reading skills (I16) or writing skills (I17).This could also be reflected by 52.0% of high achievers, who would initiate conversation in English language while communicating with others (I14).In creating structure for input and output related to English language, respondents with higher language proficiency also summarised the information in comprehending and performing the target language (I23).On the other hand, only one SILL item related to cognitive strategies showed a mixed variation in which more average achievers (58.7%) would prefer sound practising of English (I12) compared to high achievers (54.7%) and low achievers (51.4%) in learning the language.
In overcoming linguistic limitations through compensation strategies, respondents with better English language proficiency were found to use more of intelligent guessing strategies for unfamiliar words (I24) or use similar words or phrases in speaking or writing (I29).The use of gesture (I25) which showed mixed variation revealed that 48.2% of moderate achievers highly used this physical motion strategy during conversation which was the highest compared to high (48.1%)and low (35.6%)achievers.
This study also revealed that respondents with better command of English significantly had more interaction with others compared to low achievers in learning English (Table 3).This could be indicated by two SILL items under social strategies that showed positive variation and differed significantly based on the levels of language proficiency by respondents (Table 4).Respondents with better English language proficiency showed a higher percentage of "high strategy use" in cooperating with peers to practise their English (I47) and posing questions in English language (I49) in order to support their learning of English language indirectly.
In coordinating self-learning processes for English language, four metacognitive strategies-related SILL items showed positive variation and an item with mixed variation of "high strategy use".
Respondents with higher level of language proficiency would employ more of this strategy, namely to arrange and plan their English learning for maximal outcomes.This could be shown by increased percentage of respondents with better command of English in probing various ways to use English (I30), finding ways to be better English learners (I33) and seeking as many practice opportunities in English as possible (I36) (Table 4).Although SILL item related to goal setting for English improvement (I37) showed a mixed variation, percentage of high achievers with "high strategy use" still outnumbered the other two categories of learners.In the process of centering the learning of English, respondents with higher English language proficiency would direct more of their attention to English language spoken by others (I32).
Out of six SILL items related to affective strategies in learning English, two items showed significant differences of "high strategy use" based on respondents' language proficiency.The use of these two individual items of affective strategies showed mixed variation.Average achievers (65.2%) used more of self-encouragement strategies to speak in English despite of their fear in making mistakes (I40) compared to those high (63.2%)and low achievers (51.4%).Relatively, all levels of language achievers recorded low percentage of "high strategy use" in recording their feeling of learning English on their language learning diaries (I43).

Conclusion
This study revealed that learners with high language proficiency employed more strategies ranging from high to medium levels compared to moderate and low language proficiency learners, who employed all the strategies at the medium strategy range.Findings also indicated that metacognitive strategies were highly employed by all learners significantly.In addition, low achievers least preferred memory strategies whereas average and high achievers least preferred affective strategies.Further tests revealed that there were significant differences in the use of language learning strategies based on their levels of language learning proficiency.In spite of these differences, no significant differences were found between high and average achievers in the use of direct strategies even though these two categories of learners showed greater tendency to employ direct strategies.It was also reported that cognitive, compensation, social, metacognitive and affective strategies were significantly employed by learners from various levels of language proficiency, except for memory strategies which showed contrasting results.
This study also revealed that 46% of the SILL items showed significant differences based on respondents' levels of language proficiency, with the highest proportion of items from cognitive strategies, followed by metacognitive strategies, compensation strategies and affective or social strategies with similar percentage of differences.The findings from the individual items of each category of LLSs could be used as point of reference for strategy training in order to enhance the language proficiency of learners.This view was consistent with the view by Seifoori and Yaghchi (2015), who advocated that "explicit strategy training" (p.403) will enhance effective learning process where learners were exposed to how they should learn using various strategies.However, it should also be noted that not all learners were able to discern which LLSs will work out best for them in the process of learning the English language (Seifoori & Yaghchi, 2015).According to Takeuchi et al. (2007), the flexibility in applying the appropriate LLSs in a specific context will determine the learning outcomes and not the frequency of the use of LLSs (as cited in Kamalizad & Samuel, 2015, p. 3).
Furthermore, a study by Alias, Sidhu and Fook (2013) revealed that unemployed graduates who scored higher MUET band score also possessed higher general communication skills.With that in view, they suggested that follow-up steps, like revising the curriculum and creating awareness of the importance of communication skills among the graduates should be proposed by the universities to enhance the optimum level of communication skills in the workplace (Alias et. al, 2013).One of the ways was to use appropriate LLSs that could enhance the communication skills of learners by addressing the different needs of learning based on their language proficiency.In addition, students who were good in English would be able to achieve better results in their courses (Al-Qahtani, 2013).Furthermore, Razak, Ismail, Aziz, & Babikkoi (2012) voiced that effective second language learners would apply LLSs appropriately compared to the less-effective learners.They also regarded LLSs as important tools for active, self-directed involvement, which are crucial for developing communicative competence.Such view was further agreed by Huang and Nisbet (2014), who stressed that many researchers were of the view that LLSs must be "consciously deployed and carefully orchestrated in order to effective tools" (p.2).
In conclusion, future research should be geared to identify the individual LLSs based on the language proficiency of learners so that educators, trainers and other relevant parties could prepare language learning materials and curriculum to suit the needs of learners with various levels of language proficiency.Exposing learners to expand their repertoires of strategies and to use these strategies frequently and appropriately would contribute to more successful in language learning.

Figure 1 :
Figure 1: Significant Order of Strategies in Learning English by Respondents

Table 1 :
Descriptive Statistics of Strategies in Learning English based on Language Proficiency

Table 2 :
Variation of Strategies in Learning English by Respondents

Table 3 :
Variation of Strategies in Learning English based on Language Proficiency

Table 4 :
Significant Variation of SILL Items based on Language Proficiency