SYSTEM

System 35 (2007) 338352 www.elsevier.com/locate/system

The relationship between language learning strategies, prociency, age and self-ecacy beliefs: A study of language learners in Botswana
Joel Mokuedi Magogwe
a

a,*

, Rhonda Oliver

b,1

Communication and Study Skills Unit, University of Botswana, P.O. Box 402454, Gaborone, Botswana b Chair of Education Faculty of Regional Professional Studies, Edith Cowan University (South West Campus), Robertson Drive, Bunbury, WA 6230, Australia Received 9 August 2006; received in revised form 8 January 2007; accepted 24 January 2007

Abstract This research seeks to extend our current knowledge by exploring the relationship between preferred language strategies, age, prociency, and self-ecacy beliefs. Responding to the call for more replication of strategy research and for research in dierent cultural contexts, this research was undertaken in Botswana between 2002 and 2005. The adapted versions of the Oxford [Oxford, R., 1990. Language learning strategies: what every teacher should know. Newbury House, New York] Strategies Inventory for Language Learning (strategies) and the Morgan-Jinks Student Ecacy Scale [Jinks, J.L., Morgan, V.L., 1999. Childrens perceived academic self-ecacy: an inventory. Retrieved 13/9/2004, from http://www.coe.ilstu.edu/scienceed/jinks/ecacypub97.htm] (self-ecacy) instruments were used to gather this data. The results indicate that Botswana students do use a number of language learning strategies, but that they show distinct preferences for particular types of strategies. The ndings also reveal a dynamic relationship between use of language learning strategies and prociency, level of schooling (representing age dierences) and self-ecacy beliefs. These results may be used in the future to inform pedagogy and as such the outcomes from this research are important for a country where the learning of English is an important educational requirement. 2007 Elsevier Ltd. All rights reserved.
Keywords: Language learning strategies; Self-ecacy beliefs; Age; Level of education; Botswana
*

Corresponding author. Tel.: +267 3552421/3164636/71695346. E-mail addresses: magogwej@mopipi.ub.bw (J.M. Magogwe), rhonda.oliver@ecu.edu.au (R. Oliver). Tel.: +61 9780 7761; fax: +61 9780 7814.

0346-251X/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.system.2007.01.003

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

339

1. Background 1.1. English in Botswana This study was conducted in Botswana, southern Africa. In Botswana English is an ocial language used in education and commerce and Setswana is the national language spoken by 85% of the Batswana (the people of Botswana). The other 15% speak their local language(s) in addition to Setswana. English is a primary medium of instruction and a compulsory subject from Standard Three (i.e., the third year of formal schooling) onwards. However, in spite of its importance, for many in Botswana English is a second language and its use for day-to-day communication is limited. In spite of this, English is a prerequisite for further education, and candidates who enroll in a tertiary degree, with the exception of those doing science courses, must have passed English in their nal examination in senior secondary school. Despite its pivotal role in education and commerce, many Botswana students across all levels of education are not procient in the English language. They cannot speak English uently and do not perform well in the English examinations, especially at higher levels of education. Hence the purpose of this study is to explore factors that may contribute to an improvement in English language learning in Botswana. In particular, because of their apparent contribution to second language acquisition (SLA), it is the aim of this study to examine those language learning strategies used by Botswana students and the relationship between these and other aspects (e.g., age, level of education, prociency, self-ecacy) that are deemed to have an inuence on the process of acquisition. Thus the current research, being undertaken in an African context, responds to a call made by Oxford (1993) for more language strategy research with students from dierent cultural backgrounds. 2. Literature review 2.1. Language learning strategies Language strategies have long been associated with eective language learning (for example, Cohen, 1998; Green and Oxford, 1995; Hsiao and Oxford, 2002; OMalley and Chamot, 1990). With a few notable exceptions (e.g., Phillips, 1991; Green and Oxford, 1995) it generally has been claimed that good language learners use strategies more frequently, and in a greater number of situations, than do weaker/less procient learners (e.g., Ehrman and Oxford, 1990; Green and Oxford, 1995; Rubin, 1975). Rubin (1975), who along with Stern (1975) was one of the rst to undertake research in this area, describes language learning strategies as skills utilized by a learner to attain knowledge. They are techniques consciously used by learners to improve their progress in acquiring, storing, retaining, recalling, and using information in the second or foreign language (OMalley and Chamot, 1990; Oxford, 1990). The utility of language learning strategies is reected in the various ways they have been classied. For example, Oxford (1989) in her Strategies Inventory for Language Learning (SILL) prescribes six categories namely: Memory strategies, cognitive strategies, compensation strategies, metacognitive strategies, aective strategies and social strategies. However, she also cautions that there is overlap between some of the strategy categories.

340

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

Planning, for instance, can be both a metacognitive strategy, and, because it also requires reasoning, a cognitive strategy. Oxfords classication has been selected for this study because it is comprehensive, detailed and systematic (Vidal, 2002). In addition, it is reliable and valid across many cultural groups, and it links individual strategies, as well as groups of strategies, with each of the four language skill areas of listening, reading, speaking and writing (Oxford and Burry-Stock, 1995). These links are seen as being of particular relevance given the school and university settings of the current research. 2.2. Age and stages of learning or schooling Very little research has directly addressed the issue of age with respect to language learning strategies. Of that which has been conducted, Oxford found that students of different ages and stages of L2 learning use dierent strategies, with certain strategies used more by older or more advanced learners. In a study involving 348 students in a private language school in New Zealand, Griths (2003) discovered a positive correlation between level and reported frequency of language learning strategy use. Thus the current study, which was undertaken with students from various levels of education (primary, secondary and tertiary), addresses the dearth of research exploring the eect of age (given that stage of schooling generally correlates with age) on choice of language learning strategies. By undertaking research such as this, it is envisaged that educators in Botswana, and elsewhere, may be assisted in developing appropriate curricula and instructional strategies for students of dierent ages (Twyford, 1988). 2.3. Language prociency Language prociency has been consistently linked with strategy use (e.g., Green and Oxford, 1995; Khaldieh, 2000; Wharton, 2000) the general pattern being that increased prociency is linked to greater strategy use. However, there have also been results that suggest that the relationship is more complex than a simple linear correspondence between developing prociency and strategy use, and depends to a large extent on the type of strategy employed. Chen (1990), for instance, found that more procient learners actually used fewer communication strategies, though they used them more eectively than less procient students. At the same time because of the correlational nature of this type of research causality cannot be claimed (Mahlobo, 2003) and as such it cannot be determined whether the language prociency comes before, after or concurrently with strategy use (Halbach, 2000). Prociency has been determined in a multitude of ways by various researchers. Green and Oxford provide examples of the ways prociency has been determined including: selfratings (Oxford and Nyikos, 1989); language achievement tests (OMara and Lett, 1990; Phillips, 1991); entrance and placement examinations (Mullins, 1992); language course grades (Mullins, 1992); years of language study (Watanabe, 1990); and, career status (Ehrman and Oxford, 1989). In the current research, prociency has been determined on the basis of the teachers judgments about their students which were, in turn, based on the marks/grades and their personal on-balance assessment of their students given the descriptors of high prociency or good; middle prociency or fair; low prociency or poor. This was done for a variety of

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

341

reasons including the lack of appropriate English language assessments available in Botswana and because of practical considerations related to the large sample size and large geographic region from which the data were collected. 2.4. Self-ecacy beliefs Self-ecacy beliefs have been dened as: peoples judgment of their capabilities to organise and execute courses of action required to attain designated types of performances (Bandura and Schunk, 1981, p. 31). Bandura (1984) suggests that students judgments of their capability to perform academic tasks, that is, their self-ecacy beliefs, predict their capability to accomplish such tasks. Further, it is suggested that these judgments of self-ecacy mediate the eect of other inuences, such as aptitude or previous achievement, on subsequent performance. It has been observed that highly ecacious students are condent about what they can achieve; set themselves challenges and are committed to achieving them; work harder to avoid failure; are highly resilient and link failure with insucient eort or decient knowledge and skills which they believe they are capable of acquiring (Ching, 2002). Previous SLA investigations have found a strong link between increased strategy use and success (e.g., Ehrman and Oxford, 1990; Green and Oxford, 1995; Rubin, 1975), with success being measured using prociency and/or grades. This suggests that there maybe a connection between increased strategy use and higher self-ecacy beliefs. Investigations into self-ecacy and general learning strategies provide support for this hypothesis. Pajares and Schunk (2001), for example, found that students who believed they were capable of performing tasks used more cognitive and metacognitive strategies and persisted longer than those who did not. Similarly, other research ndings suggest that self-ecacy beliefs are related to self-regulated learning variables and use of learning strategies (for example see Feather, 1988; Fincham and Cain, 1986; Pape and Wang, 2003). Although Wang (2004) indicates self-ecacy beliefs have been rarely investigated in relation to SLA, the few studies that have been undertaken all suggest that a relationship does indeed exist between self-ecacy beliefs and strategy use (e.g., Purdie and Oliver, 1999; Rossiter, 2003). Given the small number of studies undertaken to date, there is clearly a need for more research in this area. This is particularly true with respect to the issue of age: The general self-ecacy belief literature indicates that self-ecacy beliefs are related not only to academic achievement but also to age and/or level of education (e.g., Pajares and Schunk, 2001). However, it is currently unknown if this is also true in relation to self-ecacy and language learning strategies. It is one aim of the current study to address this gap in the literature. 3. Research questions The specic questions addressed in this research are: 1. What types of language learning strategies do the Botswana students use? 2. What is the relationship between the language learning strategies chosen by Botswana students, and their age/level of schooling, their prociency, and their self-ecacy beliefs?

342

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

4. Method 4.1. Participants Data was collected between 2002 and 2005 from 480 students from primary schools, secondary schools, and a tertiary institution. These students were selected to represent the various ages of Botswana students. For the purpose of consistency only participants from government schools were selected. It should be noted that when compared to the other institutions used for this study, the University of Botswana is not wholly government owned and administered. At each level of education, the students were selected using a stratied random sample method; according to location, gender, and level of English language prociency (as determined by the respective teachers and lecturers). Of the 480 students who participated 168 were primary school students, 175 secondary students and 137 tertiary students. The primary students were selected from four primary schools, two located in the south and the other two in the north of the country. The secondary students were selected from four (4) schools, again two from the south and two from the north. All the tertiary students were enrolled at the University of Botswana. The background information of these participants is presented in Table 1. 4.2. Instruments A modied version of the SILL (50-item Version 7.0 for ESL/EFL) (Oxford, 1989) was used to collect information on strategies. The background questionnaire accompanying the SILL instrument was also adapted and used to collect the students demographic information. There were two versions of the questionnaire, one for both primary and secondary schools, and the other for university students (these questionnaires are available on request). All items in the questionnaires were designed for a Likert scale response using a four-interval scale of Strongly Disagree, Disagree, Agree, and Strongly Agree.

Table 1 Demographic information Primary n Gender F M Age 510 1115 1620 2125 26+ L1 Setswana English Other 80 88 2 153 11 1 1 145 1 22 % 47.6 52.4 1.2 91.1 6.5 0.6 0.6 86.3 0.6 13.1 Secondary n 87 88 0 2 173 0 0 158 2 14 % 49.7 50.3 0 1.1 98.9 0 0 90.8 1.1 8 Tertiary n 87 50 2 0 82 47 4 119 2 5 % 63.5 36.5 1.5 0 59.9 34.4 2.9 86.9 1.5 10.9

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

343

The Morgan-Jinks Student Ecacy Scale (MJSES) was used in the current research to collect self-ecacy information. It was selected because, as Jinks and Morgan (1999) indicate, the MJSES scale has undergone extensive development to assure validity and reliability. There were three versions of the questionnaire adapted to suit primary, secondary and university students. Again, the instruments were pilot tested before data collection occurred. On this basis it was determined that there was no need to translate the questionnaires. However, the four-interval Likert scale responses were adapted to Strongly Disagree, Disagree, Agree, and Strongly Agree as these descriptors are more familiar to the participants. In addition, in order to clarify the subject or language referred to, the word English was added to some of the questions in the questionnaire. 4.3. Procedure Before data collection could proceed, permission was sought from the relevant authorities both at executive/ministry and school levels, and from parents and guardians. Students were given advance notice, between one to three days, that they would be doing the questionnaires. Otherwise, the detailed process of administering the questionnaires was adhered to as closely as possible as described by the original authors. Students were advised that their responses would not aect course grades and therefore they were asked to answer honestly. 4.3.1. Analysis As a rst step in the analysis of this data, the reliability of each instrument was calculated. In this study the SILL questionnaire had an alpha reliability coecient of 0.89 for the primary version; 0.82 for the secondary; and, 0.84 for tertiary version. With respect to the MJSE Scale it had an alpha reliability coecient of 0.75 for the primary; 0.68 for the secondary; and, 0.67 for the tertiary versions of the questionnaire. Although low, these levels are still deemed to be within the acceptable range. Descriptive and inferential statistics were then used to analyze the data. Firstly, the means and standard deviations of the questionnaire data were computed. Next, one sample t-tests and one-way analysis of variance (ANOVA), repeated measures ANOVA and mixed factorial ANOVA tests were used to determine the signicance of variation in mean strategy use across the six SILL categories and the mean self-ecacy beliefs across the MJSES by level of education and prociency. To determine where the specic dierences lay Least Signicance Dierences (LSD) and Bonferroni post-hoc tests were used. Finally, analysis was undertaken to investigate the correlation (Pearson Product Moment) between self-ecacy beliefs and ESL learning strategies across prociency levels and level of education. 5. Findings 5.1. Overall use of language learning strategies From the results it can be seen that the means for the students overall strategy use, regardless of level of education, are quite similar (see Table 2). In contrast there do appear to be dierences according to the prociency of the learners, although the results suggest that in this study this was true for only those students at primary and secondary levels.

344

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

Table 2 Overall strategy use by primary, secondary and tertiary students of dierent prociency levels Prociency Primary n Good Fair Poor 58 57 54 m 2.99 2.83 2.82 SD 0.276 0.408 0.398 Secondary n 57 60 57 m 2.84 2.90 2.76 SD 0.247 0.269 0.275 Tertiary n 39 34 64 m 2.85 2.93 2.84 SD 0.245 0.292 0.232

Using a one-way ANOVA the results show that there is a signicant eect for prociency (F(2, 167) = 3.88, p = 0.23) for primary school students. The LSD post-hoc test shows that the signicant dierences are between the good and the fair students (p = 0.02) and good and poor students (p = 0.014). There is no signicant dierence between fair and poor students. Even so, the pattern of these results support previous ndings that students of high prociency use more strategies than those of low prociency, and this is true for students of younger ages. For the secondary school students, the pattern appears to be slightly dierent both to the results of the primary students, and to those reported in numerous other studies, with the fair students having the highest overall mean (M = 2.90) for strategy use, and although a one-way ANOVA shows that there is a signicant eect for prociency (F(2, 173) = 3.779, p < 0.05), the LSD post-hoc test shows that the signicant dierence is only between the fair and the poor students (p = 0.02). Therefore, the general trend is consistent with previous claims that the use of strategies increases with prociency. A similar pattern of strategy use by the secondary students also occurs in the tertiary student group. However, unlike the results for the lower level of education, the one-way ANOVA results show that there is no signicant eect for prociency (F(2, 136) = 1.474, p = 0.233) with respect to strategy use at the tertiary level. Although this nding is contrary to previous research, when considered in relation to the results for the younger groups of students it does suggest a dynamic relationship between level of education and prociency in learner strategy use. Overall these strategy use results clearly indicate that primary, secondary and tertiary Botswana students do indeed use language learning strategies. As such they are consistent with the results of other language learning strategy studies which show that L2 learners from dierent cultural backgrounds use language learning strategies in an apparent attempt to become eective learners of English language (for example, Carson and Longhini, 2002; Cohen, 1990; Hsiao and Oxford, 2002; OMalley and Chamot, 1990; Rubin, 1975). Further, it would also seem that the general pattern emergent from the literature (e.g., Green and Oxford, 1995; Khaldieh, 2000; Wharton, 2000) that more procient learners use more strategies, generally holds true for this cohort of learners in Botswana. 5.2. Categories of language learning strategies The dierent pattern of strategy use for the three age groups also applies in relation to the categories of strategies, with each age group showing preference in a dierent order. For example, at the primary school level the mean score of social strategy use is greater, in turn, than that of metacognitive, cognitive, aective, memory and compensation strategies (see Table 3). The results of a one-way repeated measures ANOVA show that there is

J.M. Magogwe, R. Oliver / System 35 (2007) 338352 Table 3 Strategy use by primary, secondary and tertiary students Strategy Primary n Metacognitive Social Cognitive Aective Memory Compensation 168 168 168 168 168 168 m 3.06 3.21 2.91 2.79 2.76 2.51 SD 0.513 0.500 0.430 0.493 0.508 0.551 Secondary n 174 173 174 174 174 172 m 3.22 3.05 2.91 2.63 2.57 2.45 SD 0.403 0.449 0.337 0.461 0.384 0.498 Tertiary n 137 137 137 137 137 137 m 3.26 2.83 2.93 2.73 2.69 2.54

345

SD 0.353 0.460 0.348 0.436 0.365 0.433

a signicant eect for strategy category (F(5, 835) = 71.66, p < 0.001). Further, the Bonferroni post-hoc test shows that at this level of education there is a signicant dierence between the mean preferred use of most categories of strategies except between the memory strategies and cognitive strategies, and between memory strategies and aective strategies. At the secondary level a slightly dierent pattern of use occurs. Specically, in terms of mean scores the most preferred strategies are of a metacognitive kind. The next most preferred type of strategy for this group is social (which, by contrast, the primary school students prefer most), then cognitive, aective, memory and compensation strategies. Again a oneway repeated measures ANOVA test shows a signicant eect for strategy category (F(5, 850) = 122.347, p < 0.001) with these secondary students. The Bonferroni post-hoc test shows a signicant dierence between most of the categories except between the memory and compensation, memory and aective strategies, and compensation and aective strategies. At the tertiary level, yet a dierent pattern of preference emerges. For this group whilst again the metacognitive strategies have the highest mean, they are followed by cognitive, social, aective, memory, and compensation strategies. The results of a one-way repeated measures ANOVA show that there was a signicant eect for strategy category (F(5, 675) = 71.457, p < 0.001). The Bonferroni tests showed a signicant dierence between all other categories except between memory and aective strategies, cognitive and social strategies, and aective and social strategies. Next is presented the use of various categories of strategies (e.g., social, cognitive) by good, fair and poor prociency students in the dierent levels of education (see Table 4). The emerging picture is that at primary level all the students, regardless of prociency, indicated a preference for more social, metacognitive and cognitive strategies and fewer compensation and aective strategies. At the same time, however, the younger students of higher prociency levels indicated that they used more social and metacognitive strategies than did the fair and poor prociency students. A mixed factorial ANOVA test shows a signicant eect for the strategy category (F(5, 825) = 73.687, p < 0.001), but not for prociency (F(2, 165) = 3.216, f < 1), although there is a signicant interaction between prociency and strategy categories (F(10, 825) = 4.947, p < 0.001). Therefore, for this level of education it would seem that prociency inuences the use of strategies of particular types. At the secondary school level good and fair students indicated a greater preference for metacognitive, social and cognitive and compensation strategies than did students deemed to be of a poor prociency level. The mixed factorial ANOVA test results show that there is a signicant eect for strategy category (F(5, 840) = 122.962, p < 0.001), but again there

346

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

Table 4 Strategy use of primary, secondary and tertiary students by prociency Strategy Primary n Metacognitive Good Fair Poor Good Fair Poor Good Fair Poor Good Fair Poor Good Fair Poor Good Fair Poor 58 57 54 58 57 54 58 57 54 58 57 54 58 57 54 58 57 54 m 3.25 3.02 2.88 3.35 3.20 3.07 3.03 2.83 2.87 2.85 2.69 2.72 2.89 2.80 2.69 2.45 2.41 2.68 SD 0.383 0.499 0.547 0.419 0.548 0.540 0.357 0.446 0.464 0.483 0.500 0.534 0.462 0.508 0.498 0.507 0.563 0.555 Secondary n 57 60 57 57 60 57 57 60 57 57 60 57 57 60 57 57 60 57 m 3.24 3.25 3.17 3.08 3.09 2.99 2.92 2.99 2.82 2.54 2.75 2.58 2.59 2.58 2.53 2.50 2.55 2.30 SD 0.444 0.401 0.363 0.391 0.471 0.481 0.319 0.318 0.357 0.431 0.461 0.467 0.341 0.423 0.385 0.450 0.551 0.456 Tertiary n 39 34 64 38 34 64 39 34 64 39 34 64 39 34 64 39 34 64 m 3.28 3.32 3.21 2.86 2.95 2.75 2.88 2.98 2.94 2.76 2.77 2.68 2.63 2.73 2.70 2.50 2.63 2.52 SD 0.385 0.349 0.333 0.432 0.384 0.503 0.330 0.353 0.356 0.451 0.425 0.435 0.366 0.410 0.340 0.449 0.467 0.404

Social

Cognitive

Aective

Memory

Compensation

is no signicant eect for prociency, nor this time is there a signicant interaction between prociency and strategy categories (F(10, 840) = 1.318, F < 1). In other words, at this level of education the prociency level of students did not signicantly inuence the preference for strategies from particular categories even though overall the more procient students (fair and good) used more strategies than did poor students. At the tertiary level all students indicated a greater preference for metacognitive, cognitive and social strategies than they did for other strategies. However, more good and fair students recorded preference for metacognitive strategies than did the poor prociency students. Even so, although the results from a mixed factorial ANOVA show that there was a signicant eect for strategy category (F(5, 665) = 67.146, p < 0.001), there was no signicant eect for prociency (F(1, 133) = 1.858, F < 1), nor was there a signicant interaction eect for category and prociency (F(10, 665) = 0.732, F < 1). Thus these ndings support an observation made by Wharton (2000) that the types of strategies used depend on the kind of learners and setting in which learning occurs. Specically, the preference for particular types of strategies was inuenced by prociency at the primary school level, but not at the secondary or tertiary level. Further, when compared to learners in other cultural settings it is apparent that the combination of strategies preferred by the Botswana students is not the same as that as for other learners. For example, Chang (1991) showed that the most preferred strategies of the 50 Chinese ESL students in his study were compensation strategies and the least preferred were aective strategies, whereas in this study compensation strategies were the least preferred across all levels of education. The results across the three levels of schooling also suggest that particular strategies may be developmentally acquired. For example, both the secondary and tertiary level

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

347

students preferred metacognitive strategies, whereas the primary school students preferred social strategies. This may occur because students at higher levels of education are more independent learners and metacognitive strategies best match this characteristic. Even so, social strategies were preferred more often than other strategies by students from all levels of education in Botswana. This may reect a cultural approach to learning in this particular African context, or it may be that the social strategies have synergies with the English language teaching approach now being used in that country (where there has been a shift toward developing sociolinguistic and communicative competencies). At the same time, however, there are many similarities between this data and that collected in other contexts. For example the students in this study indicated a low level of preference for aective strategies. This is similar to the results of studies such as those by Chamot and Kupper (1989) and Goh and Kwah (1997). It may also be that like students in Oxfords (1993) study, students in Botswana are largely unaware of the potential of aective strategies. In summary it is apparent that the students in this study used all six categories of language learning strategies, regardless of educational level and prociency (at the secondary and tertiary levels). At the same time it does seem that in the Botswana context some categories are preferred over others. Specically, metacognitive, social and cognitive strategies were more preferred than aective, memory and compensation strategies across all levels of education, suggesting the possibility that these may be culturally determined strategies. 5.3. Self-ecacy beliefs As previously indicated, an adapted form of the MJSES questionnaire was used to ascertain the participants self-ecacy beliefs. The following scale was used to score the students responses to this: 1 represents strongly disagree, 2 disagree, 3 agree and 4 strongly agree. Therefore means above 2 represent positive self-ecacy beliefs, with higher results equating to stronger and more positive beliefs. The overall self-ecacy means were positive and quite similar across the three levels of education. At the primary level the overall self-ecacy mean was 2.79 (SD = 0.24, n = 168). At the secondary level, although the overall mean (M = 2.60; SD = 0.22, n = 173) was lower than that for the primary school students, it still shows positive self-ecacy beliefs. The overall mean of the MJSES results for the tertiary students was 2.68 (SD = 0.25, n = 136). Not surprisingly when prociency is also considered, the results show small dierences between the groups, with the pattern being the higher the prociency, the higher the selfecacy beliefs. Thus it can be seen that good students generally scored the highest mean followed by fair students and then poor students (see Table 5). However, the one-way
Table 5 Self-ecacy beliefs of primary, secondary and tertiary students by prociency Prociency Primary n Good Fair Poor Total 58 56 54 168 M 2.84 2.78 2.74 2.79 SD 0.208 0.338 0.351 0.305 Secondary n 57 60 56 173 M 2.65 2.60 2.56 2.60 SD 0.209 0.219 0.228 0.220 Tertiary n 39 33 64 136 M 2.69 2.69 2.67 2.68 SD 0.249 0.259 0.250 0.250

348

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

ANOVA results showed that there were no signicant dierences between the means of the three prociency groups, whether for primary students (F(2, 167) = 1.531, p = 0.219), for secondary students (F(2, 172) = 2.023, p = 0.135), or for tertiary students (F(2, 133) = 0.117, p = 0.890). 5.4. The relationship between self-ecacy beliefs and language learning strategies To examine whether or not relationships exist between the students strategy use and self-ecacy beliefs, the overall results of SILL and MJSES were statistically compared using Pearson Product Moment correlations. Firstly this was done for students at dierent levels of education and then in terms of their prociency. As shown in Table 6, for primary school there is a moderate, positive (r = 0.588) and signicant (p < 0.001) correlation between self-ecacy beliefs and overall use of strategies. Specically the results show that an increase in self-ecacy beliefs of primary school students is related to an increase in their use of strategies. However, this relationship is not strong. At the secondary school level the relationship between self-ecacy beliefs and overall use of strategies is moderate (0.435), positive and signicant (p < 0.001). Finally, at the tertiary level a weak, positive (0.297) and signicant (p < 0.001) relationship exists between self-ecacy beliefs and overall use of strategies. 5.5. The relationship between self-ecacy beliefs, strategies, level of education and prociency An analysis was made of the relationship between self-ecacy beliefs, language learning strategy use and prociency for students at the dierent levels of education. The results suggest that the factors of age (as represented here by levels of education) and prociency combine in complex ways with regard to their relationship with self-ecacy beliefs and use of language learning strategies (see Table 7). The emerging pattern is that as the level of

Table 6 Relationship between self-ecacy beliefs and overall use of strategies for primary, secondary and tertiary students Strategy n Pearson R signicance Primary 168 0.558 0.001 Secondary 172 0.435 0.001 Tertiary 136 0.297 0.001

Table 7 Correlation between self-ecacy and strategies for dierent prociency groups of primary, secondary and tertiary students Prociency Primary n Good Fair Poor 58 56 54 Pearson R 0.367 0.482 0.699 Signicance 0.005 0.001 0.001 Secondary n 57 59 56 Pearson R 0.280 0.432 0.557 Signicance 0.035 0.001 0.001 Tertiary n 39 33 64 Pearson R 0.044 0.504 0.323 Signicance 0.791 0.003 0.009

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

349

prociency increases, the correlation between self-ecacy beliefs and use of strategies decreases, although this varied slightly for the students at dierent levels of education. In terms of prociency, the primary school results show a weak (0.367), positive, but not signicant (p = 0.005) relationship between self-ecacy beliefs and use of strategies for good students. For those students of fair prociency the relationship is moderate, positive (0.482) and signicant (p < 0.01). Comparatively, there is a strong, positive (0.699) and signicant (p < 0.001) correlation for poor prociency students. Previous research (Pajares and Schunk, 2001) has suggested that high self-ecacy beliefs are associated with high achievement, and similarly high use of strategies has also been related to the qualities of good language learning and according to these results this appears to be the case in Botswana, at least with respect to primary school students. At secondary school level, the results mirror those found for primary school students as they show a weak (0.280), positive but not signicant (p = 0.035) relationship between selfecacy beliefs and use of strategies for good students. On the other hand there is moderate, positive (0.432) and signicant (p < 0.01) correlation for fair students. Comparatively, there is a strong, positive (0.557) and signicant (p < 0.001) correlation for poor prociency students. It can be seen that, as for primary school students, the higher the students reported use of strategies the higher their self-ecacy beliefs, although given the lack of signicance for good students this can be best described as a trend. Further, the correlation weakens as prociency increases, a result similar to that of the primary school students. At the tertiary level there is a weak, positive (0.044) but not signicant (p = 0.791) relationship between self-ecacy beliefs and use of strategies for good students. Comparatively, there is a strong, positive (0.504) but again not signicant (p = 0.003) correlation for fair students. Similarly, the correlation for poor prociency students is weak, positive (0.323) but not signicant (p = 0.009) (see Table 7). Therefore, although the pattern of results is similar to that of the primary and secondary school students, none of the correlations were signicant. Therefore it is apparent that there is an interplay of relationships between the factors of self-ecacy, strategy use and prociency, and level of school. These relationships, whilst at times not strong, are dynamic and certainly worthy of much further research. 6. Limitations of this study One of the limitations of this study is that the number of schools from which data were collected is small, hence making it dicult to generalise the ndings of this research to the entire population of students in Botswana. However, it is important to point out that the students used in this study share important common attributes such as that, in the respective levels, they all belong to government or government aided schools, they belong to the same level of education, their ages do not vary a great deal and that they use the same English curriculum. A more signicant limitation in this study is a methodological one: Teachers and lecturers were asked to select good, fair and poor students either by using marks or their knowledge of the students performance in English. It should be noted, however, that there may have been some variation in the procedures used by the teachers and lecturers to select the students. Although this was based on previous strategy research, a standardised test may have provided data that allowed for more reliable comparisons.

350

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

7. Conclusion As anticipated, the results of this research are consistent with the general ndings of previous language learning strategy studies (see Dreyer, 1992; Green and Oxford, 1995; OMalley and Chamot, 1990; Taguchi, 2002) and at the same time provide new evidence for strategy use, in this case in the Botswana context. Like previous research, this study found more overall use of language learning strategies among more procient learners than less procient ones, and this was consistent across all levels of education. However, the ndings of this research also suggested more complex patterns of use than have appeared in earlier studies and they also provide indication that there is a relationship between type of strategy use and successful language learning, but that this is mediated by a number of factors, including self-ecacy beliefs. With respect to the type of strategies used, the ndings of the current research show some dierences according to level of school and, in the primary school, according to prociency. It is also apparent that particular strategies may be culturally more appropriate, and therefore preferred, or it may that the educational experience of Botswana students leads them to prefer some strategies (e.g., social strategies) over others. It would also seem that the choice of particular strategies may be determined, at least in part, by the age of the learners. For example, in this study secondary and tertiary students were more likely to choose metacognitive strategies than were their primary school counterparts, perhaps because of their level of cognitive development. The ndings of the current research show that Botswana students across all levels of education are moderately ecacious about their learning of the English language, although not consistently so. With regard to age the self-ecacy belief results were mixed, sometimes favouring younger learners and at other times older learners. The results of this study also showed that there was a positive and signicant but weak relationship between self-ecacy beliefs and use of overall language learning strategies across all prociency levels (although this relationship is not strong). Despite this association not being strong, the results clearly indicate that there is still room for a great deal more research relating to strategy use and self-ecacy, and in turn, these in relation to age and prociency. For example, the fact that poor prociency learners with high self-ecacy, use strategies more often (which could, in time, lead to improved prociency) is an area worthy of further exploration. In terms of English language learning in Botswana, these results may be used in the future to inform pedagogy. For example, strategies should be incorporated into the curriculum and the students should be explicitly taught how to use the strategies. This is particularly pertinent given that English is an important educational requirement. References
Bandura, A., 1984. Recycling misconceptions of perceived self-ecacy. Cognitive Therapy and Research 8, 231 255. Bandura, A., Schunk, D.H., 1981. Cultivating competence, self-ecacy, and intrinsic interest through proximal self-motivation. The Journal of Personality and Social Psychology 41, 586598. Carson, J.G., Longhini, A., 2002. Focusing on learning styles and strategies: a diary study in an immersion setting. Language Learning 52 (2), 401438. Chamot, A., Kupper, L., 1989. Learning strategies in foreign language instruction. Foreign Language Annals 22 (1), 1324.

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

351

Chang, S.F., 1991. A study of language learning behaviours of Chinese students at the University of Georgia and the relation of those behaviours to oral prociency and other factors. University of Georgia, Athens, GA. Chen, S.Q., 1990. A study of communication strategies in interlanguage production by Chinese EFL learners. Language Learning 40, 155187. Ching, L.C., 2002. Strategy and self-regulation instruction as contributors to Improving students cognitive model in an ESL programme. English For Specic Purposes 13, 261289. Cohen, E.G., 1990. Teaching in multiculturally heterogeneous classrooms: ndings from a model program. McGill Journal of Education 26 (1), 723. Cohen, A.D., 1998. Strategies in Learning and Using a Second Language. Longman, Essex, UK. Dreyer, C., 1992. Learner Variables as Predictors of ESL Prociency. Potchefstroom University, South Africa. Ehrman, M., Oxford, R., 1989. Eects of sex dierences, career choice, and psychological type on adult language learning strategies. The Modern Language Journal 72 (iii), 253265. Ehrman, M., Oxford, R., 1990. Adult language learning strategies in an intensive training setting. The Modern Language Journal 74 (3), 311327. Feather, N.T., 1988. Values, valences, and course enrollment: testing the role of personal values within an expectancy-valence framework. Journal of Educational Psychology 80, 381391. Fincham, F., Cain, K., 1986. Learned helplessness in humans: a developmental analysis. Developmental Review 6, 138156. Goh, C., Kwah, P.F., 1997. Chinese ESL students learning strategies: a look at frequency, prociency and gender. Hong Kong Journal of Applied Linguistics 2, 3953. Green, M., Oxford, R., 1995. A closer look at learning strategies, L2 prociency and gender. TESOL Quarterly 29, 261297. Griths, C., 2003. Patterns of language learning strategy use. System 31 (3), 367383. Halbach, A., 2000. Finding out about students learning strategies by looking at their diaries: a case study. System 28, 8596. Hsiao, T., Oxford, R., 2002. Comparing theories of language learning strategies: a conrmatory factor analysis. The Modern Language Journal 86 (3), 368383. Jinks, J.L., Morgan, V.L., 1999. Childrens perceived academic self-ecacy: an inventory. Retrieved 13/9/2004, from http://www.coe.ilstu.edu/scienceed/jinks/ecacypub97.htm. Khaldieh, S.A., 2000. Learning strategies and writing processes of procient vs. less-procient learners of Arabic. Foreign Language Annals 33 (5), 522533. Mahlobo, E.B., 2003. The relationship between standardised test performance and language learning strategies in English as s second language: a case study. Journal For Language Teaching 37 (2), 164176. Mullins, P. (1992). Successful English language strategies of students enrolled in the Faculty of Arts, Chulalongkorn University, Bangkok Thailand. United States International University, San Diego, CA. OMalley, J.M., Chamot, A., 1990. Learning Strategies in Second Language Acquisition. CUP, Cambridge. OMara, F., Lett, J., 1990. Foreign Language Learning and Retention: Interim Results of the Language Skill Change Project. Advanced Technology, Inc., Reston, VA. Oxford, R., 1989. Use of language learning strategies: a synthesis of studies with implications for strategy training. System 17, 235247. Oxford, R., 1990. Language Learning Strategies: What Every Teacher Should Know. Newbury House, New York. Oxford, R., 1993. Research on second language learning strategies. Annual Review of Applied Linguistics 13, 175187. Oxford, R., Burry-Stock, J.A., 1995. Assessing the use of language learning strategies worldwide with the ESL/ EFL version of the strategy inventory for language learning (SILL). System 23 (1), 123. Oxford, R., Nyikos, M., 1989. Variables aecting choice of language learning strategies by university students. The Modern Language Journal 73 (ii), 291300. Pajares, F., Schunk, D.H., 2001. Self-ecacy beliefs and school success: Self-ecacy, self-concept, and school achievement. In: Perception. Ablex Publishing, London, pp. 239266. Pape, S.J., Wang, C., 2003. Middle school childrens strategic behavior: classication and relation to academic achievement and mathematical problem-solving. Instructional Science 31, 419449. Phillips, V., 1991. A look at learner strategy use and ESL prociency. CATESOL Journal, 5767. Purdie, N., Oliver, R., 1999. Language learning strategies used by bilingual school-aged children. System 27, 375 388.

352

J.M. Magogwe, R. Oliver / System 35 (2007) 338352

Rossiter, M.J., 2003. The eects of aective strategy training in the ESL classroom. Retrieved 13/9, 2004, from http:/www-writing.berkely.edu//tesl-ej/ej26/az.html. Rubin, J., 1975. What the good language learner can teach us. TESOL Quarterly 9, 4151. Stern, H.H., 1975. What can we learn from the good language learner? Canadian Modern Language Review 31, 304318. Taguchi, T., 2002. Learner factors aecting the use of learning strategies in cross-cultural contexts. Prospect 17 (2), 1831. Twyford, C.W., 1988. Age-related factors in second language acquisition. Focus, 2. Vidal, R.J., 2002. Is there a correlation between reported language learning strategy use, actual strategy use and achievement? Linguagem & Ensino 5 (1), 4373. Wang, C., 2004. Self-regulated learning strategies and self-ecacy beliefs of children learning English as a second language. Unpublished Ph.D., Ohio State University, Educational Theory and Practice. Watanabe, Y., 1990. External variables aecting learning strategies of Japanese EFL learners: eects of entrance examination, years spent at college/university, and staying overseas. Unpublished masters thesis, Lancaster University, Lancaster, UK. Wharton, G., 2000. Language learning strategy use of bilingual foreign language learners in Singapore. Language Learning 50 (2), 203243.