AUTHOR: Stephen Bremner

TITLE: Language Learning Strategies and Language Proficiency: Investigating the Relationship in Hong
Kong
SOURCE: Canadian Modern Language Review 55 no4 490-514 Je 1999

The magazine publisher is the copyright holder of this article and it is reproduced with permission. Further
reproduction of this article in violation of the copyright is prohibited.

ABSTRACT
This article reports on a survey of the language learning strategies used by a group of Hong Kong learners. The
aims of the study were to investigate levels of strategy use among the group, and to examine levels of
association between strategy use and language proficiency. The SILL questionnaire (Strategies Inventory of
Language Learning) by Oxford (1990, pp. 293-300) was used. SILL consists of six categories of strategies:
memory, cognitive, compensation, metacognitive, affective, and social. The results showed that compensation
and metacognitive strategies were the most used, while affective and memory strategies were the least used.
Previous examinations of the nature of the relationship between strategy use and proficiency, and ways of
measuring this are discussed. In this study, it was found that there was significant variation in proficiency in
relation to eleven out of a possible fifty strategies. Of these, nine were in the cognitive category, one in the
compensation category, and one in the social category. The article concludes by questioning the appropriateness
of using the SILL and proficiency measure in tandem as a way of establishing a clear relationship between
strategy use and proficiency, and suggests directions that might be pursued in language learning strategy
research. Cet article offre un compte rendu d'une ้tude portant sur les strat้gies d'apprentissage des langues
utilis้es par un groupe d'apprenants เ Hong Kong. Les buts vis้s par cette ้tude ้taient de relever le niveau
d'utilisation des strat้gies parmi les membres du groupe, et ensuite d'analyser le niveau d'association entre
l'utilisation des strat้gies et la comp้tence linguistique. On a utilis้ le questionnaire de l'inventaire des strat้gies de
l'apprentissage des langues (SILL), mis au point par Oxford (1990, pp. 293-300). Les strat้gies y sont divis้es en
six cat้gories: m้morielle, cognitive, compensatoire, m้tacognitive, affective et sociale. Les r้sultats montrent que
les strat้gies compensatoire et m้tacognitive ้taient les plus souvent utilis้es, tandis que les strat้gies affective et
m้morielle l'้taient le moins. On discute des ้tudes pr้c้dentes portant sur la nature du lien entre utilisation de
strat้gies et comp้tence linguistique, ainsi que des facons de mesurer ce lien. Dans cette ้tude, on a trouv้ qu'il
existait une variation importante du rapport entre la comp้tence et la strat้gie dans onze des cinquante strat้gies
possibles. Parmi ces onze strat้gies, neuf se situaient dans la cat้gorie cognitive, une dans la cat้gorie
compensatoire et une dans la cat้gorie sociale. L'article conclut en questionnant la pertinence de l'utilisation du
SILL conjointement avec la mesure de la comp้tence comme m้thode permettant d'้tablir un lien net entre
utilisation de strat้gie et comp้tence, proposant de nouvelles voies de recherche sur les strat้gies de
l'apprentissage des langues.

INTRODUCTION
The idea that there may be a set of strategies used by language learners to help them learn language has been
with us for some time. From early examples of research such as the studies carried out by Rubin (1975) and
Stern (1975), to taxonomies of strategies like that drawn up by Oxford (1990), to theories of language
acquisition which incorporate strategies (O'Malley & Chamot, 1990), much work has been done in attempting
to identify what might be good language learning strategies, and in trying to establish a relationship between
these and successful language learning.
This paper examines the links between language learning strategy use and L2 proficiency of a group of
learners in Hong Kong. The term 'proficiency' here can be interpreted in the sense used by Bachman to refer in
general to 'knowledge, competence or ability in the use of a language, irrespective of how, where, or under what
conditions it has been acquired' (1990, p. 16). Specifically, the paper looks at two questions:
1. What type of learning strategies do the group of Hong Kong learners under investigation report themselves
as using?
2. What associations, if any, exist between the levels of strategy use and proficiency among the group? (In
looking at the latter, I also discuss the relationship between strategy use and proficiency, and the extent to which
they can be treated as separate: in other words, are they interlinked features of learner performance, or separate
factors, one the cause of the other?)
In order to address these questions, I discuss some of the studies that have examined the nature of language
learning strategies, in particular those which have looked at possible links between strategy use and proficiency,
and then I present the findings from a study carried out with a group of students at the City University of Hong
Kong.
BACKGROUND -- IDENTIFYING AND CATEGORISING STRATEGIES
Early research into language learning strategies was concerned with attempting to establish what good
language learning strategies might be. While no explicit claims were made about links between strategies and
success, the titles of these early articles implied a relationship: 'What the "Good Language Learner" can Teach
Us' (Rubin, 1975), and 'What Can We Learn From the Good Language Learner?' (Stern, 1975). Researchers
were hoping to identify strategies used by successful learners with the idea that they might be transferred to less
successful learners. As Horwitz says: 'The ultimate purpose of studying learner strategies is, of course, an
applied one; researchers and teachers hope to determine which strategies are most effective and help students
adopt more productive learning procedures' (1987, p. 126). Notable studies carried out since these early two
include Naiman, Frohlich, Stern, and Todesco's (1978) study, Rubin's (1981) study, and the work done by
O'Malley, Chamot, Stewner-Manzanares, Kupper, and Russo (1985).
There is considerable debate as to appropriate ways of defining language learning strategies, and no strong
consensus as yet (see Ellis, 1994, and Gu, 1996, for a discussion of this issue). The terms which have been used
to describe strategies (e.g. technique, behaviour, operation, action) and to account for their purpose (to acquire
knowledge, to regulate learning, to make learning more effective) vary, but they have much in common. Oxford
defines strategies in general terms as 'steps taken by learners to enhance their own learning' (1990, p. 1).
As well as the various ways of defining strategies, there are also different ways of categorising those that
have been identified. Based on research that took place during the 1980s, O'Malley and Chamot (1990), for
example, outline a scheme which includes cognitive, metacognitive, and social/affective strategies. Cognitive
strategies work with information in ways that enhance learning; metacognitive strategies are described as
'higher order executive skills' that could involve the planning, monitoring, or evaluation of an activity, and
social/affective strategies entail 'interaction with another person or ideational control over affect' (1990, pp. 44-
45). Oxford (1990) has also produced a classification system which takes account of research conducted over
the preceding years, including much of her own work. This will be looked at in greater detail later.
A wide range of methods of data collection has been used: observation and interviews (Naiman et al., 1978);
observations, student self-report and diaries (Rubin, 1981); self-report through questionnaires (Politzer, 1983;
Politzer & McGroarty, 1985; Oxford, 1990); and interviews (Wenden, 1987).
Thus the situation that has evolved is one where there is no clear agreement on a definition for strategies, and
where there is a variety of classification schemes and elicitation methods. The existing schemes of classification
have areas in common, but, as is noted by Oxford and Ehrmann, 'The proliferation of strategy systems has
caused problems for those researchers who believe it is important to compare results across studies' (1995, p.
363). However, one instrument for eliciting levels of strategy use has gained currency through being
administered in a variety of learning environments: Oxford's (1990) Strategy Inventory for Language Learning
(SILL). This has been developed from Oxford's fairly exhaustive list of strategies, described by Ellis as 'perhaps
the most comprehensive classification of learning strategies to date' (1994, p. 539). According to Green and
Oxford (1995), studies using SILL have involved around 8,000 students in different parts of the world.
This research has used the SILL for the reason that it has been used extensively and thus allows for
comparisons to be made; the study which is the principal point of comparison for this paper is the research
carried out by Green and Oxford (1995), using the same instrument. In their study, the relationships between
strategy use, L2 proficiency and gender were examined using a population of students in a university in Puerto
Rico. The results and analysis procedures are discussed later in this paper. I shall also refer to a study conducted
by Goh and Kwah (1997) in which they looked at the strategy use of tertiary-level students from PRC learning
English in Singapore. This too used the SILL, again investigating relationships between strategy use, gender,
and proficiency.

THE RELATIONSHIP BETWEEN PROFICIENCY AND STRATEGY USE

Various studies have investigated the relationship between proficiency and strategy use. Politzer and
McGroarty (1985), for example, looked at the relationship between a range of 'good learning behaviours,'
measured using a questionnaire, and gain scores on an intensive course. They reported mixed results: while the
gain scores did not relate to their categories of strategy use as a whole (classroom behaviours, individual study
behaviours, and social interaction behaviours outside the classroom), there were certain individual items which
showed significant associations with certain of their proficiency measures. They also found differences in
reported strategy use among the two main groups, Asians and Hispanics. In concluding, however, the authors
offer a number of caveats regarding the use of self-report data, and say that, 'Results indicate that caution in
prescribing good learning behaviors is warranted' (1985, p. 103).
Abraham and Vann (1987; Vann and Abraham 1990), in two separate studies, looked at the language learning
strategies employed by both successful and unsuccessful learners. These distinctions were made by measuring
'the relative speed with which they moved through an intensive English program' (1990, p. 179). They found
that unsuccessful learners were using strategies generally considered as useful, and often the same ones as those
employed by the successful learners: the difference lay in the degree of flexibility the learners showed when
choosing strategies, and how appropriately they were applied to the given situation. The findings from these
studies raise a question mark over the idea that successful learners use a larger repertoire of strategies, and use
them more frequently.
This is slightly different from the picture painted by Green and Oxford (1995): they say that in studies in a
variety of settings investigating the relationship between proficiency and language learning strategy use,
'students who were better in their language performance generally reported higher levels of overall strategy use
and frequent use of a greater number of strategy categories' (1995, p. 265).
Findings such as those mentioned by Green and Oxford might indicate an association between reported
strategy use and proficiency, but the exact nature of this association, particularly the issue of causality, is a
subject of some debate. Skehan (1989) and Rees-Miller (1993) among others have pointed out that the existence
of correlation between the two does not necessarily suggest causality in a particular direction. McIntyre (1994)
has attempted to unravel the relationship between the two variables. On the one hand he stresses a need for
caution when looking at studies which suggest that more proficient students make better use of strategies: 'This
might be interpreted to mean that either proficiency influences the choice of strategies or that strategy choice is
simply a sign of proficiency level' (1994, p. 188). However, in answer to his own question as to whether
strategy use results from or leads to increased proficiency, he is rather less cautious: 'The answer, undoubtedly,
is BOTH' (1994, p. 189; emphasis in original). The case made for this statement is not a strong one: the idea that
strategies are both the causes and outcomes of improved language proficiency needs much more investigation
before such a confident conclusion can be warranted.
A similar argument to McIntyre's, although offered more tentatively, comes from Green and Oxford (1995):
in their Puerto Rico study, they found that about a third of the individual strategies were used more frequently
by the more successful learners, almost all of them involving active use of the target language. Although they
concede that this is not sufficient evidence of causality, they nevertheless suggest that a causal relationship
exists here between strategy use and proficiency level, and that 'this relationship is best visualised not as a one-
way arrow leading from cause to effect, but rather as an ascending spiral in which active use strategies help
students attain higher proficiency, which in turn makes it more likely that students will select these active use
strategies' (1995, p. 288). This may be a plausible theory, but there is no evidence for it beyond the significant
levels of variation in use of certain strategies that they found among different groups of students. It could
equally be argued that strategies do not contribute to proficiency, but are simply features of it; in other words,
only by reaching a certain level will a student be likely to use a given strategy. This point is made by Skehan:
'One can ... argue that learner strategies do not determine proficiency, but are permitted by it' (1989, p. 97;
emphasis in original).
The notion that strategy use and proficiency are both causes and outcomes of each other, locked in a mutual
relationship, complicates the picture: certainly the assumptions that a researcher makes about the direction of
causality in this relationship between proficiency and strategy use will have implications for the way in which
their study is conducted, and this issue is discussed in the methods section of the paper.

THE CURRENT STUDY

The purpose of the present study was to investigate the strategy use of a group of students at the City
University of Hong Kong, and to focus on the relationship between the subjects' strategy use and one variable,
namely their language proficiency. This compares with studies that have looked at other variables in
conjunction with strategy use, such as gender and proficiency (Goh & Kwah 1997; Green & Oxford, 1995), or
major subject studied at university (Oxford & Nyikos, 1989).

SUBJECTS
The subjects (n=149, 113 female, 36 male) were all studying a language and communication skills course at
the City University of Hong Kong, part of a BA in Primary Education; as this was an in-service course, all of
them were already working as primary teachers. There was a wide divergence in age, from mid-twenties to mid-
fifties. The L1 of the group was Cantonese. While all had undergone compulsory instruction in English at
school, the ultimate length of language study varied, as did proficiency. Students were placed into classes for
logistical reasons (i.e., according to when they could attend) rather than on the basis of proficiency level. The
course lasted twenty weeks: thirty hours of instruction and forty-five hours of self-study. The syllabus for all
groups was the same.

INSTRUMENTS
The instrument used for collecting data on strategy use was Oxford's (1990, pp. 293-300) Strategy Inventory
for Language Learning (50-item Version 7.0 for ESL/EFL). The SILL is a self-scoring survey and the structure
is based on Oxford's classification system, whereby strategies are grouped into two types: direct (i.e., strategies
which directly involve the target language) and indirect. These in turn are divided into categories: memory,
cognitive, and compensation (direct); metacognitive, affective, and social (indirect).
 Memory strategies relate to the storing and retrieval of information (e.g., 'I use new English words in a
sentence so I can remember them'); cognitive strategies, although varied, are described by Oxford as being
'unified by a common function: manipulation or transformation of the target language by the learner' (1990, p.
43) (e.g., 'I use the English words I know in different ways'). Meanwhile compensation strategies 'enable
learners to use the new language for either comprehension or production despite limitations in knowledge' (p.
47) (e.g., 'To understand unfamiliar English words I make guesses'). As for the indirect group, metacognitive
strategies 'allow learners to control their own cognition' (p. 135) (e.g., 'I look for people to talk to in English'),
affective strategies are concerned with the regulation of feelings and attitudes (e.g., 'I try to relax whenever I
feel afraid of using English'), and social strategies are those which take account of the fact that language is a
form of social behaviour, involving communication with other people (e.g., 'I practise English with other
students').
Each item in the survey is a statement saying 'I do ...' (e.g., 'I pay attention when someone is speaking') and
students respond on a 5-point Likert scale ranging from 1 ('Never or almost never true of me') to 5 ('Always or
almost always true of me'). An English version of the questionnaire was administered roughly halfway through
the course. Green and Oxford quote reliability of various forms of SILL, using Cronbach's alpha for internal
consistency, as .93-.98 (1995, p. 264). Reliability of SILL for this population, again using Cronbach's alpha, was
.92.
In order to determine the students' proficiency in relation to each other, the following tasks were administered
in the first and second half of the twenty-week course: two spoken tasks, two written tasks, and two discrete-
item language tests. Scores from these were totalled and converted into a percentage. The breakdown of
assignments is as follows:

Task Percentage of total score

(First half of course)
Speaking: interactive task 15%
Writing: short essay (in-class) 15%
Discrete-item language test 20%
(Second half of course)
Speaking: mini-presentation 15%
Writing: short essay (outside class) 15%
Discrete-item language test 20%

Scoring for the written and spoken tasks was subjective, based on criteria agreed by the two instructors when
the tasks were set; scoring for the discrete-item language tests was mostly objective. While an objective test of
overall proficiency might have been preferable, the assessment procedures for the course were already in place,
and it was felt that the range of tasks set for students was sufficiently comprehensive. A statistical check for
reliability using Cronbach's alpha was run, calculating the intercorrelations between the tasks set in the first half
and those set in the second half, with the result r = .9005. For the different sections -- speaking, writing and
discrete item tests -- the results were .8543, .7339, and .8305 respectively. This meets Skehan's requirements for
test reliability: he says that a reliability coefficient of 0.7 or above is acceptably high (1989, p. 12).

DATA COLLECTION AND ANALYSIS PROCEDURES

Green and Oxford's (1995) study and this research both use analysis of variance (ANOVA) to determine
significant variation. Are the assumptions for the use of ANOVA met in this study? Brown (1988) states it is a
requirement that the scores in each group are normally distributed and that -- if the group sizes are very unequal
-- the variances of the scores are approximately equal. A check of the data found that both assumptions were
met.
As mentioned earlier, assumptions made about the direction of causality among variables are likely to
influence the analysis procedures. The nomination of variables, then, raises an important question: by setting the
independent variable, which the use of ANOVA necessitates, is one positing an implicit direction of causality?
According to Oxford and Nyikos (1989), the answer is yes: 'causality is by definition involved in the use of the
ANOVA technique; indeed, that is why researchers use it (when possible) instead of less explanatory techniques
like correlation' (1989, p. 295).
In Green and Oxford's (1995) study, which was the original point of comparison for the current study, the
independent variables are gender and proficiency, and the dependent variable is strategy use. Why this
configuration of dependent and independent variables? Clearly gender has to be an independent variable, as the
direction of causality in this relationship can only be in one direction. In the case of proficiency, however, if one
assumes that the goal of learning strategy research is to establish whether strategy use has a positive effect on
the enhancement of proficiency, it would seem to be more logical to set strategy use as an independent variable.
Given these differing positions on the setting of variables, it was decided to conduct two analyses: the first
was the approach used by Green and Oxford, i.e., using ANOVA with proficiency as the independent variable
and strategy use as the dependent variable. This was carried out to allow for direct comparison with data from
previous research. The second analysis was carried out in line with the concerns outlined earlier, thus strategy
use was nominated as the independent variable, and proficiency as the dependent variable.

TRADITIONAL APPROACH -- PROFICIENCY AS INDEPENDENT VARIABLE

The subjects were divided into three groups of roughly equal numbers on the basis of proficiency scores.

Group Number(n) Range of scores(% )

1 (Low) 48 48-70
2 (Medium) 51 70.5-78.5
3 (High) 50 79-94

Differences in mean strategy use in the six SILL categories (the dependent variables) in relation to
proficiency by group (the independent variable) were determined using a one-way analysis of variance
(ANOVA). The Scheffe post-hoc test was used to see where any significant differences lay. To determine
significance throughout the study, a standard of p < .05 was used.

ALTERNATIVE APPROACH -- PROFICIENCY AS DEPENDENT VARIABLE

This approach looked at the relationship between individual strategies and proficiency, with the latter as the
dependent variable. Three groups of students were established for each strategy, characterised as low, medium,
and high strategy users. Low consisted of responses 1 and 2 ('never or almost never true of me' and 'usually not
true of me'), medium was 3 ('somewhat true of me'), and high corresponded to responses 4 and 5 ('usually true
of me' and 'always or almost always true of me'). Again, significant variation in proficiency by individual
strategy use was determined using a one-way analysis of variance (ANOVA) in conjunction with the Scheffe
post-hoc test.

THE FINDINGS
Table 1 shows the mean strategy use in each of these categories for the entire group. Although there are
differences in level of use by strategy group, all means for the six strategy categories fell within the range of 2.5
to 3.4, which is defined by Oxford (1990) as medium use. The compensation category had the highest mean,
followed by metacognitive, cognitive, social, memory and affective.
A summary of the ANOVA results for the six categories of strategy, where proficiency is the independent
variable, is shown in Table 2.
These results show significant relationships with proficiency for three categories of strategy, with two
(cognitive and compensation) representing positive variation, i.e., more use by students with higher proficiency,
and one (affective) representing negative variation.
ANOVA was run for each of the individual strategies, and it was found that there was significant positive
variation in proficiency in relation to eleven strategies, and mixed variation in relation to five strategies. The
strategies are listed in Table 3, and a summary of the ANOVA results for the strategies is shown in Table 4.
Of the strategies which showed significant association with proficiency, nine fall into the cognitive category
(11, 12, 13, 15, 17, 18, 19, 20, 22), one into the compensation category (27), and one into the social category
(49).

DISCUSSION

LEVELS OF STRATEGY USE

As can be seen from Table 1, the Hong Kong students reported using compensation strategies more than any
other strategy, and this was followed by metacognitive strategies. It is interesting to compare these findings
with those from other studies looking at strategy use among Chinese learners. Two studies looking at students
from Taiwan and the PRC (Klassen, 1994; Yang, 1994, cited in Oxford & Burry-Stock, 1995) also reported
compensation strategies as being the most frequently used, falling in the high range of use (i.e., above 3.4)
while all other strategies fell in the medium range. Goh and Kwah's (1997) survey of students from the PRC
learning English in Singapore also reported high use of compensation strategies (3.46), with only metacognitive
strategies being used more frequently (3.54).
Similarly, all three of these studies reported memory strategies as having the lowest frequency, while this
Hong Kong study has memory strategies as second lowest in reported frequency. Goh and Kwah point out that
their findings seem 'to contradict commonly accepted accounts of the learning strategies of Chinese learners'
(1997, p. 47). They speculate that while their students might have a tradition of memorisation, they probably did
not know about the specific techniques in the SILL, such as using rhymes or connecting the sounds of new
words to an image or picture, a theory which was confirmed when they interviewed a number of students.
Further research into the memorisation techniques used by Chinese learners would be necessary to establish
whether the SILL was an appropriate instrument for investigating their strategy use in this area.
The picture that emerges is that the learners in this study report similar levels of strategy use, based on
Oxford's categories, to groups of Chinese learners in other environments. But how do the strategies most
frequently reported by the Hong Kong learners relate to their levels of proficiency?

VARIATION IN USE OF THE SIX CATEGORIES OF STRATEGIES (PROFICIENCY AS

INDEPENDENT VARIABLE)
Significant differences by proficiency level were found in the use of three of the broad strategy categories.
Two of these, cognitive and compensation, showed positive variation, indicating higher use of these strategies
among higher proficiency levels. Goh and Kwah (1997) also found significant positive variation in the use of
the same two categories, while Green and Oxford (1995) found these two and the metacognitive and social
categories to show significant variation. It is difficult to speculate about the relationship between levels of
proficiency and frequency of cognitive strategy use, as it appears that many of the cognitive strategies (e.g., 'I
use the English words I know in different ways') could either be contributors to the acquisition of proficiency or,
alternatively, be made more possible by increased proficiency. As for compensation strategies, intuitively it
would seem that they might contribute to the learning process, but it is equally plausible (if we see proficiency
as afecting strategy use) that learners with lower levels of proficiency may not have the confidence or base of
language knowledge to make informed guesses or make up new words: compensation strategies might be more
readily available to learners who have reached higher levels of proficiency.
The third category, affective, showed negative variation i.e. higher use of these strategies among lower levels
of proficiency. As this form of analysis, using proficiency level as the independent variable, is implicitly
investigating the effect of proficiency on strategy use, one might conclude that as learners become more
proficient, they have less need of affective strategies, or, taking the argument a step further, that these are not
really strategies for learning, but simply features which exist among low level learners. This was the least
frequently reported category for all groups, though, so it is perhaps difficult to conclude too much from this
finding.
Much of the above discussion of the findings with regard to variation in categories of strategy use is
influenced by concerns about the setting of dependent and independent variables: examination of the findings
where strategy use is the independent variable is more extensive.

LANGUAGE PROFICIENCY AND STRATEGY USE (STRATEGY USE AS INDEPENDENT

VARIABLE)
In this study eleven strategies all together showed significant association with proficiency level; this
compares with seventeen in the 1995 Green and Oxford study. Eight of the eleven are among their seventeen
(only items 18 and 19 did not appear on their list). While the data analysis procedures in this study differed
from theirs, as discussed above, it is nevertheless interesting to note that these eight were seen as showing
significant association in both studies. For their own findings, Green and Oxford observe that the majority of
the strategies used more frequently by more successful learners involved active language use, and go on to
suggest that 'a crucial role in L2 learning appears to be played by strategies involving active use of the target
language' (1995, p. 291).
Do the results from this survey support their claim? Certainly most of the eleven strategies could be loosely
defined as 'involving active use of the target language,' but the exact role that these play cannot be determined
from these data. The findings support Green and Oxford's claim only if one accepts that strategy use has an
effect on proficiency, but the issue of direction of causality in relation to these two variables has not been
resolved.
Goh and Kwah, however, looking for variation among learners of different levels of proficiency, appear to be
clear about where the direction of causality lies: 'Results from two-way ANOVA showed that the proficiency
level of the students had a significant influence on the use of two categories of learning strategies: cognitive and
compensation' (1997, p. 48). The implication of this seems to be that by raising the proficiency level of
students, we would be helping them to use strategies more frequently. As suggested earlier, encouraging
strategy use so as to increase proficiency would appear to be a more logical and fruitful aim.
This Hong Kong study is concerned with examining the effect of strategy use on proficiency. The findings
indicate that higher use of the eleven strategies listed in Table 3 is associated with higher levels of proficiency.
The implicit direction assumed in using ANOVA is that use of these strategies is a cause rather than an outcome.
However, the argument that higher use of these strategies is the outcome of increased proficiency still cannot be
ruled out. Indeed, it is quite plausible that students with higher levels of proficiency are quite likely to use
words in more ways, to watch more films in English, to write more notes, to ask more questions, and that their
higher level of proficiency actually enables them to do these things more readily. Certainly on the basis of the
findings in this and other studies, there is no clear indication of causality in one particular direction, rather there
are significant associations between proficiency level and use of certain strategies. Other techniques would be
necessary to establish causal relationships.

IMPLICATIONS OF THE FINDINGS

There are three possible ways of looking at strategies and their relationship with proficiency: the first is to see
them as the outcomes of proficiency, in which case there is no need to investigate them, rather to look at what
helps students acquire proficiency; the second is to see them as having a uni-directional causal role in increasing
proficiency, but there is no strong evidence for this as yet; the third is to accept the view of McIntyre (1994) or
Green and Oxford (1995), as outlined earlier, that the relationship between the two is mutual, that causality is
bi-directional.
If we assume the second or third options, then what can be said about the strategies found in this and other
studies to show significant association (in either direction) with proficiency levels? What do these strategies
represent for the teacher? In other words, how can the students' interpretations of the SILL items be converted
into teachable techniques to be passed on to other learners?

LEARNERS' INTERPRETATION OF STRATEGIES

To answer this question requires finding out exactly how learners interpret items in strategy questionnaires,
i.e., asking how accurate a picture of student strategy use we can get from instruments like the SILL. Certainly
we must approach data from self-report questionnaires with a certain amount of caution, and this is necessitated
by considerations such as the possible range of interpretation of the questionnaire items, the cultural context in
which the questionnaire has been administered, and the reliability of the reports that each learner provides.
The statements that constitute the different items students respond to may be vague and open to a range of
interpretations. For example: 'I pay attention when someone is speaking' (item 32). Who is the someone? A
teacher in a strict classroom environment? A friend with whom you have to speak English as your only common
means of communication? A fellow student in a group activity? Very often the SILL is translated into the L1 of
the respondents, but it is difficult to see how such potential ambiguity or vagueness could be avoided in any
language. Respondents in LoCastro's (1994) study were cited as criticising 'the lack of contextualisation of
some items ... Overwhelmingly, they suggested it depends on the situation and the people' (1994, p. 412).
The cultural context will also affect students' interpretations. It is interesting to note, for example, that in this
study there was a significant association between proficiency level and stated use of item 19 ('I look for words
in my language that are similar to new words in English'). If this were a group of French learners, it would be
natural to assume that they were looking for words in French which were similar to English words, of which
there are many instances. But in a Chinese context what could this mean?
Similarly, it cannot be assumed that a particular strategy will be useful in any cultural context. What may be
beneficial in one cultural environment may be considered to have a different effect in another. Take item 32 ('I
ask questions in English'): in many learning contexts it is seen as a positive form of participation. But this
depends on one's view of what constitutes participation: in one environment it might mean offering ideas and
asking questions, but in another, such as the PRC, it might mean not asking questions; questions might be
construed as being disruptive, disrespectful, and not in the interests of the group. As Rees-Miller says: 'Can it be
said that a student who does not ask questions in class or correct his peers is inactive?' (1993, p. 684). Thus both
the interpretation of items and the views of what might have a positive effect on learning will vary from context
to context.
As for the issue of the reliability of the students' responses, Politzer and McGroarty advise caution for
researchers using self-report data concerning learning behaviours, saying that they can 'reflect general
intelligence, a desire to give the "right" answer or to please the teacher and so on' (1985, p. 118). See also
Cohen (1987) for a general discussion of using verbal reports in research.
There is a range of factors, then, which can affect the picture of strategy use that emerges from elicitation
instruments such as the SILL. In this study, given the decontextualised nature of the questionnaire items and the
problems of interpretation that this might cause, many of the eleven strategies with significant levels of
association with proficiency level may appear rather vague to a teacher, because they lack a context, and might
seem to be little more than a set of broad practice behaviours. Do they represent behaviurs that can be translated
into teachable techniques? Item 18 ('I first skim an English passage then go back and read it carefully') and item
27 ('I read English without looking up every word') could be passed on as strategies for reading; but item 12 ('I
practise the sounds of English'), item 13 ('I use English words I know in different ways'), and item 49 ('I ask
questions in English'), for example, are too general to be anything but exhortations to practise more.
What I am suggesting is that the research methodology used here, the combination of the SILL and
proficiency scores, is perhaps not in itself sufficient to provide practical, useable insights into the relationship
between strategy use and proficiency level. Oxford and Green (1995) discuss what the SILL is appropriate for,
and point out that no single data-generating technique can serve every purpose. The SILL, they say, provides 'a
good general picture of strategy use' (1995, p. 167), as it does for this group of Hong Kong learners. However,
when we analyse learners' levels of strategy use in conjunction with their proficiency level, we are simply
providing a general idea of the associations between the two at one moment in time. We are not measuring the
effect of one on the other. If we are to learn more about this relationship, we need to use a much narrower focus.
Instruments such as the SILL could be a starting point for such research, providing a general idea of which
strategies have significant levels of association with proficiency. It would then be necessary to establish what
these strategies actually mean to students in their particular learning situation, and then to find a way of
converting them into teachable techniques, the effect of which could be measured over a period of time with
different groups of learners. The SILL provides a snapshot, but only by using more longitudinal approaches will
we get an idea of the possible effects that language learning strategies have on proficiency.

CONCLUSION
In terms of frequency of use of different categories of strategy, this study has shown results similar to other
studies conducted among Chinese-speaking students in different parts of the region - the PRC, Taiwan, and
Singapore - in that compensation and cognitive strategies are reported as being used the most frequently, while
memory strategies (as represented in the SILL) are among the least frequently used. Further studies would be
necessary to establish how widespread a phenomenon this is among Chinese speakers, and how different their
patterns of strategy use are from those of other groups in other cultures.
As for the relationship between levels of strategy use and proficiency, the study indicates significant levels of
association for eleven strategies, nine of which are from the cognitive category, and as in Green and Oxford's
(1995) study, many of these are seen to involve active practice.
However, the study does raise questions about the effectiveness of this method of investigating the
relationship between strategy use and proficiency level: firstly, there is the problem of attempting to establish a
direction of causality within the relationship; analysing this relationship using the SILL and proficiency scores
alone does not provide enough information in this respect, because it does not measure the effect of one factor
on the other over a period of time. Secondly, strategy use as reported by the SILL alone does not give a full
picture of what these strategies represent for students, with the resulting problem that teachers cannot easily
translate them into teachable strategies for use in the learning process.
The ultimate goal of any research trying to establish connections between strategy use and learning success
should be to establish the extent to which strategy use contributes to the learning process. Research which looks
at the effect of proficiency on strategy use seems to be going down a different track. If strategies are not causes
but features of proficiency, then they are not worth investigating - they are simply outcomes of increased
proficiency, an increase that has to be accounted for in other ways. If, on the other hand, they are contributory
factors towards increased proficiency, then they do warrant further investigation. However, in their general,
decontextualised form as items on a questionnaire, those that showed significant association in this study do not
represent much more than an encouragement to practise more, and it is hardly a great advance to suggest that
practice helps.
While the SILL can provide a broad idea of overall strategy use, the study of the effect of strategy use on
proficiency perhaps does not require an instrument as comprehensive in its scope as this: what would be more
appropriate is an approach which investigates the effect of very specific strategies on localised aspects of
proficiency, in specific contexts, over a period of time.
ADDED MATERIAL
Stephen Bremner has been working in the Department of English at the City University of Hong Kong since
1996. He has taught in the Middle East, Eastern Europe, and the People's Republic of China. His current
research interests are ESP, in particular workplace writing, and learner strategies.
This research was supported by the Educational Discourse/TESL Research Group, Department of English,
City University of Hong Kong. Thanks also to Dr Matt Peacock for his help with statistical analysis, and to the
three anonymous reviewers for some helpful comments on an earlier version of the paper.
TABLE 1 Mean strategy use in each of the six strategy groups

Strategy group Mean Standard Deviation Rank order of usage

A Memory 2.85 .48 5
B Cognitive 2.97 .55 3
C Compensation 3.36 .52 1
D Metacognitive 3.12 .70 2
E Affective 2.76 .52 6
F Social 2.91 .63 4

TABLE 2 Summary of variation in use of strategy categories (independent variable: proficiency)

Dependent
Variable (SILL Group 1 Group 2 Group 3 F Significance
category) M SD M SD M SD ratio level Comments
Memory 2.95 .53 2.77 .46 2.84 .44 n.s.
Cognitive 2.78 .55 2.94 .51 3.17 .53 6.94 p < .0013 G3>G1
Compensation 3.13 .45 3.43 .46 3.51 .57 8.16 p < .0004 G2>G1 G3>G1
Metacognitive 3.09 .67 3.00 .67 3.26 .73 n.s.
Affective 2.90 .50 2.78 .53 2.59 .50 4.75 p < .0100 G1>G3
Social 2.97 .68 2.80 .65 2.96 .57 n.s.

TABLE 3 Strategies which showed significant association with proficiency level

Positive variation
11. I try to talk like native English speakers.
12. I practise the sounds of English.
13. I use English words I know in different ways.
15. I watch English TV shows spoken in English or go to movies spoken in
English.
17. I write notes, messages, letters or reports in English.
18. I first skim an English passage (read over the passage quickly) then go
back and read carefully.
19. I look for words in my own language that are similar to new words in
English.
20. I try to find patterns in English.
22. I try not to translate word-for-word.
27. I read English without looking up every word.
49. I ask questions in English.
Mixed variation
23. I make summaries of information that I hear or read in English.
24. To understand unfamiliar words in English, I make guesses.
26. I make up new words if I do not know the right ones in English.
38. I think about my progress in learning English
44. I talk to someone else about how I feel when I am learning English.

Note: (The numbers next to each strategy are those from the questionnaire--a full list of the fifty strategies
can be found in Appendix 1).
TABLE 4 Summary of variation in proficiency

Positive variation
Proficiency score (as percentage) by strategy use (low use, medium use, high use)
SILL Low (G1) Medium (G2) High (G3) F Significance
Item M SD M SD M SD ratio level Comments
11 71.18 9.9 74.73 8.84 77.79 8.2 6.3 p < .0024 G3>G1
12 68.62 10.58 73.69 8.17 77.65 9.1 9.1 p < .0002 G3>G1
13 71.07 9.9 74.66 8.12 78.70 10.17 6.3 p < .0024 G3>G1
15 71.58 9.7 74.62 8.79 76.87 9.4 3.6 p < .0312 G3>G1
17 72.10 9.4 74.36 8.20 82.81 8.6 10.64 p < .0000 G3>G2 G3>G1
18 69.90 9.6 73.22 9.61 76.53 8.6 5.9 p < .0035 G3>G1
19 70.60 10.62 73.49 8.21 76.73 9.4 4.8 p < .0100 G3>G1
20 71.06 10.05 73.08 8.49 76.81 9.6 4.3 p < .0156 G3>G1
22 69.56 9.8 73.95 9.25 78.04 7.9 8.6 p < .0003 G3>G1
27 70.41 8.0 72.85 9.09 77.83 9.4 8.8 p < .0003 G3>G2 G3>G1
49 71.73 9.5 74.05 8.89 79.64 8.4 7.2 p < .0011 G3>G2 G3>G1
Mixed variation
Proficiency score (as percentage) by strategy use (low use, medium use, high use)
SILL Low (G1) Medium (G2) High (G3) F Significance
Item M SD M SD M SD ratio level Comments
23 73.89 10.08 72.01 8.47 77.5 8.80 3.9 p < .0226 G3>G2
24 75.00 9.00 71.04 9.06 75.49 9.43 3.4 p < .0345 G3>G2
26 74.72 10.02 71.97 9.56 77.72 6.56 4 p < .0200 G3>G2
38 74.22 9.9 72.27 8.94 78.05 8.92 5 p < .0079 G3>G2
44 77.23 8.9 68.87 7.92 71.74 9.33 12.87 p < .0000 G1>G2 G1>G3

REFERENCES
Abraham, R.G., & Vann, R.J. (1987). Strategies of two language learners: A case study. In A.Wenden & J.
Rubin (Eds.), Learner strategies in language learning (pp. 85-102), Englewood Cliffs, NJ: Prentice Hall.
Bachman, L. (1990). Fundamental considerations in language testing. Oxford: Oxford University Press.
Brown, J.D. (1988). Understanding research in second language learning. Cambridge: Cambridge University
Press.
Cohen, A.D. (1987). Using verbal reports in research on language learning. In C. Faerch & G. Kasper (Eds.),
Introspection in second language research (pp. 82-95). Clevedon, UK: Multilingual Matters.
Ellis, R. (1994). The study of second language acquisition. London: Oxford University Press.
Goh, C., & Kwah, P.F. (1997). Chinese ESL students' learning strategies: A look at frequency, proficiency and
gender. Hong Kong Journal of Applied Linguistics, 2, 39-53.
Green, J.M., & Oxford, R.L. (1995). A closer look at learning strategies, L2 proficiency, and gender. TESOL
Quarterly, 29, 261-297.
Gu, P.Y. (1996). Robin Hood in SLA: What has the learning strategy researcher taught us? Asian Journal of
English Language Teaching, 6, 1-29.
Horwitz, E.K. (1987). Surveying student beliefs about language learning. In A. Wenden & J. Rubin (Eds.),
Learner strategies in language learning (pp. 119-129). Englewood Cliffs, NJ: Prentice Hall.
Klassen, J. (1994) The language learning strategies of freshman English students in Taiwan: a case study.
Unpublished master's thesis, California State University at Chico.
LoCastro, V. (1994). Learning strategies and learning environments. TESOL Quarterly, 28, 409-414.
McIntyre, P.D. (1994). Toward a social psychological model of strategy use. Foreign Language Annals, 27,
185-195.
Naiman, N., Fr ๖ hlich, M., Stern, H.H., & Todesco, A. (1978). The good language learner. Toronto: Ontario
Institute for Studies in Education.
O'Malley, J.M., & Chamot, A.U. (1990). Learning strategies in second language acquisition. Cambridge,
England: Cambridge University Press.
O'Malley, J.M., Chamot, A.U., Stewner-Manzanares, G., Kupper, L., & Russo, R.P. (1985). Learning
strategies used by beginning and intermediate ESL students. Language Learning, 35, 21-46.
Oxford, R.L. (1990). Language learning strategies: What every teacher should know. Boston: Heinle &
Heinle.
Oxford, R.L., & Burry-Stock, J. (1995). Assessing the use of language learning strategies worldwide with the
ESL/EFL version with the strategy inventory of language learning (SILL). System, 23, 1-23.
Oxford, R.L., & Ehrman, M.E. (1995). Adults' language learning strategies in an intensive foreign language
program in the United States. System, 23, 359-386
Oxford, R.L., & Green, J.M. (1995). Making sense of learning strategy assessment: toward a higher standard
of research accuracy (Comments on Virginia LoCastro's 'Learning strategies and learning environments').
TESOL Quarterly, 29, 166-174.
Oxford, R. L., & Nyikos, M. (1989). Variables affecting choice of language learning strategies by university
students. Modern Language Journal, 73, 291-300.
Politzer, R.L., (1983). An exploratory study of self-reported language learning behaviors and their relation to
achievement. Studies in Second Language Acquisition, 6, 54-65.
Politzer, R.L., & McGroarty, M. (1985). An exploratory study of learning behaviors and their relationship to
gains in linguistic and communicative competence. TESOL Quarterly, 19, 103-123.
Rees-Miller, J. (1993). A critical appraisal of learner training: Theoretical bases and teaching implications.
TESOL Quarterly, 27, 679-689.
Rubin, J. (1975). What the 'good language learner' can teach us. TESOL Quarterly, 9, 41-51.
Rubin, J. (1981). Study of cognitive processes in second language learning. Applied Linguistics, 2, 117-131.
Skehan, P. (1989). Individual differences in second-language learning. London: Edward Arnold.
Stern, H.H. (1975). What can we learn from the good language learner? The Canadian Modern Language
Review, 31, 304-318.
Vann, R.J., & Abraham, R.G. (1990). Strategies of unsuccessful learners. TESOL Quarterly, 24, 177-198.
Wenden, A.L. (1987) How to be a successful language learner: Insights and prescriptions from L2 learners. In
A. Wenden & J. Rubin (Eds.), Learner strategies in language learning (pp. 103-117). Englewood Cliffs, NJ:
Prentice Hall.
Yang, N.D. (1994). An investigation of Taiwanese college students' use of English learning strategies.
Research report, National Taiwan University, Taiwan.
APPENDIX 1
Strategies questionnaire (Oxford, 1989)
Strategy Inventory for Language Learning (SILL)
This form of the Strategy Inventory for Language Learning (SILL) is for students of English as a second or
foreign language. You will find statements about learning English. Please read each one and write the response
(1, 2, 3, 4 or 5) that tells HOW TRUE OF YOU THE STATEMENT IS in the space next to the statement.
1. Never or almost never true of me.
2. Usually not true of me.
3. Somewhat true of me.
4. Usually true of me.
5. Always or almost always true of me.
NEVER OR ALMOST NEVER TRUE OF ME means that the statement is very rarely true of you.
USUALLY NOT TRUE OF ME means that the statement is true less than half the time.
SOMEWHAT TRUE OF ME means that the statement is true of you about half the time.
USUALLY TRUE OF ME means that the statement is true more than half the time.
ALWAYS OR ALMOST ALWAYS TRUE OF ME means that the statement is true of you almost always.
Answer in terms of how well the statement describes you. Do not answer how you think you should be, or
what other people do. There are no right or wrong answers to these statements. Work as quickly as you can
without being careless. This usually takes about 20-30 minutes to complete. If you have any questions, let the
teacher know immediately.

PART A
1. I think of relationships between what I already know and new things I learn in English.
2. I use new English words in a sentence so I can remember them.
3. I connect the sound of a new English word and an image or picture of the word to help me remember the
word.
4. I remember a new English word by making a mental picture of a situation in which the word might be
used.
5. I use rhymes to remember new English words.
6. I use flashcards to remember new English words.
7. I physically act out new English words.
8. I review English lessons often.
9. I remember new English words or phrases by remembering their location on the page, on the board, or on a
street sign.

PART B
10. I say or write new English words several times.
11. I try to talk like native English speakers.
12. I practise the sounds of English.
13. I use the English words I know in different ways.
14. I start conversations in English.
15. I watch English language TV shows spoken in English or go to movies spoken in English.
16. I read for pleasure in English.
17. I write notes, messages, letters or reports in English.
18. I first skim an English passage (read over the passage quickly) then go back and read carefully.
19. I look for words in my own language that are similar to new words in English.
20. I try to find patterns in English.
21. I find the meaning of an English word by dividing it into parts that I understand.
22. I try not to translate word-for-word.
23. I make summaries of information that I hear or read in English.

PART C
24. To understand unfamiliar English words, I make guesses.
25. When I can't think of a word during a conversation in English, I use gestures.
26. I make up new words if I do not know the right ones in English.
27. I read English without looking up every new word.
28. I try to guess what the other person will say next in English.
29. If I can't think of an English word, I use a word or phrase that means the same thing.
PART D
30. I try to find as many ways as I can to use my English.
31. I notice my English mistakes and use that information to help me do better.
32. I pay attention when someone is speaking English.
33. I try to find out how to be a better learner of English.
34. I plan my schedule so I will have enough time to study English.
35. I look for people I can talk to in English.
36. I look for opportunities to read as much as possible in English.
37. I have clear goals for improving my English skills.
38. I think about my progress in learning English.

PART E
39. I try to relax whenever I feel afraid of using English.
40. I encourage myself to speak English even when I am afraid of making a mistake.
41. I give myself a reward or treat when I do well in English.
42. I notice if I am tense or nervous when I am studying or using English.
43. I write down my feelings in a language learning diary.
44. I talk to someone else about how I feel when I am learning English.

PART F
45. If I do not understand something in English, I ask the other person to slow down or say it again.
46. I ask English speakers to correct me when I talk.
47. I practise English with other students.
48. I ask for help from English speakers.
49. I ask questions in English.
50. I try to learn about the culture of English speakers.

APPENDIX 2
Summary of variation in proficiency for all strategies
Proficiency score (as percentage) by strategy use (low use, medium use, high use)

SILL Low (G1) Medium (G2) High (G3) F Significance Com-

item M SD n M SD n M SD n ratio level ments
1 73.64 8.92 22 72.89 8.96 70 76.06 10.01 57 n.s.
2 75.78 8.50 46 73.62 10.01 57 73.38 9.57 46 n.s.
3 75.61 8.92 32 73.32 8.88 51 74.23 10.11 66 n.s.
4 73.18 8.16 34 75.04 9.17 60 73.95 10.46 55 n.s.
5 76.15 8.81 58 72.99 9.44 47 72.98 9.96 44 n.s.
6 74.95 9.30 103 72.20 9.30 35 73.77 10.92 11 n.s.
7 74.81 9.05 88 73.77 9.44 45 72.19 11.52 16 n.s.
8 72.85 9.60 42 74.24 8.97 79 76.20 10.40 28 n.s.
9 75.18 8.12 56 72.08 10.20 55 75.88 9.71 38 n.s.
10 73.30 9.88 47 74.25 8.25 51 75.03 10.19 51 n.s.
11 71.18 9.91 56 74.73 8.84 53 77.79 8.24 40 6.27 p < .0024 G3>G1
12 68.62 10.58 26 73.69 8.17 70 77.65 9.07 53 9.12 p < .0002 G3>G1
13 71.07 9.86 49 74.66 8.12 73 78.70 10.17 27 6.29 p < .0024 G3>G1
14 73.72 9.40 79 74.33 9.19 48 75.73 10.27 22 n.s.
15 71.58 9.74 48 74.62 8.79 63 76.87 9.43 38 3.55 p < .0312 G3>G1
16 73.17 9.70 69 73.97 9.24 52 77.25 8.73 28 n.s.
17 72.10 9.35 77 74.36 8.20 54 82.81 8.64 18 10.64 p < .0000 G3>G2
G3>G1
18 69.90 9.61 29 73.22 9.61 46 76.53 8.63 74 5.89 p < .0035 G3>G1
19 70.60 10.62 30 73.49 8.21 59 76.73 9.36 60 4.75 p < .0100 G3>G1
20 71.06 10.05 25 73.08 8.49 65 76.81 9.64 59 4.28 p < .0156 G3>G1
21 71.21 9.64 17 74.11 8.38 74 75.23 10.53 58 n.s.
22 69.56 9.80 33 73.95 9.25 71 78.04 7.86 45 8.56 p < .0003 G3>G1
23 73.89 10.08 61 72.01 8.47 49 77.50 8.80 39 3.89 p < .0226 G3>G2
24 75.00 8.97 7 71.04 9.06 42 75.50 9.38 100 3.45 p < .0345 G3>G2
25 71.68 11.75 11 72.69 9.42 54 75.52 9.01 84 n.s.
26 74.72 10.02 59 71.97 9.56 60 77.72 6.56 30 4.02 p < .0200 G3>G2
27 70.41 8.01 37 72.85 9.09 53 77.83 9.41 59 8.75 p < .0003 G3>G2
28 72.64 9.35 54 74.73 9.31 59 75.74 9.65 36 n.s.
29 71.07 12.00 7 70.79 8.55 26 75.17 9.32 116 n.s.
30 73.01 9.24 39 73.08 8.99 69 77.27 9.85 41 n.s.
31 70.42 8.12 12 12.49 8.97 62 76.25 9.62 75 n.s.
32 71.72 11.47 9 71.76 8.85 35 75.25 9.33 105 n.s.
33 73.96 8.76 12 71.69 9.77 44 75.44 9.20 93 n.s.
34 73.72 8.84 56 73.64 9.70 66 76.63 9.91 27 n.s.
35 73.10 9.64 70 74.57 8.98 43 75.96 9.48 36 n.s.
36 73.62 9.53 59 73.84 8.65 53 75.70 10.38 37 n.s.
37 73.37 8.88 68 73.87 9.86 48 76.45 9.79 33 n.s.
38 74.22 9.92 36 72.27 8.94 75 78.05 8.92 38 5.01 p < .0079 G3>G2
39 71.79 10.55 19 75.19 9.00 74 73.75 9.57 56 n.s.
40 73.67 8.59 27 72.54 8.83 51 75.63 10.03 71 n.s.
41 75.98 9.45 82 71.62 8.70 41 72.73 9.59 26 n.s.
42 78.06 8.90 27 73.51 8.95 59 73.23 9.81 63 n.s.
43 74.58 9.72 130 71.60 7.49 15 72.12 3.42 4 n.s.
44 77.23 8.89 86 68.88 7.92 36 71.74 9.34 27 12.88 p < .0000 G1>G2
G1>G3
45 73.92 9.97 20 73.39 8.89 36 74.60 9.59 93 n.s.
46 76.00 9.23 61 74.15 9.26 48 71.58 9.52 40 n.s.
47 74.68 9.13 59 73.09 9.20 59 75.47 10.46 31 n.s.
48 73.99 8.04 56 73.92 9.99 52 74.89 10.58 41 n.s.
49 71.73 9.53 57 74.05 8.89 64 79.64 8.42 28 7.19 p < .0011 G3>G2
G3>G1
50 74.42 8.57 56 73.68 8.83 62 74.92 12.00 31 n.s.