Language Learning 53:4, December 2003, pp.

649–702

A Closer Look at Gender and Strategy Use

in L2 Reading

Aek Phakiti
University of Melbourne

This study examines gender differences in cognitive

and metacognitive strategy use in the context of an
English as a foreign language reading comprehension
test. Three hundred eighty-four Thai university students
took a multiple-choice reading comprehension test, then
completed a questionnaire on their strategy use. Gender
differences were analyzed using multivariate analysis of
variance. Males and females did not differ in their reading
comprehension performance and their use of cognitive
strategies. Unexpectedly, males reported significantly
higher use of metacognitive strategies than females.
Within the same achievement groups (highly successful,
moderately successful, and unsuccessful), however, there
were no gender differences in either reading performance
or use of cognitive and metacognitive strategies. The article
discusses the implications for future gender-based
research.

In the past decades, considerable advances have been made

in understanding the nature of second language (L2) reading,
and these changes have influenced how L2 reading has been

The following people played a crucial role during the process of completing
the present article: Paul Gruba, Brian Lynch, Tim McNamara, Carsten
Röver, Nick Ellis, Kathleen Bardovi-Harlig, Paul Russell, and the
anonymous Language Learning reviewers.
Correspondence concerning this article should be addressed to Aek
Phakiti, Department of Linguistics and Applied Linguistics, University of
Melbourne, Victoria, 3010, Australia. Internet: [email protected]

649
650 Language Learning Vol. 53, No. 4

taught, learned, and assessed (see Alderson, 2000; Grabe, 1999).

Successful reading comprehension is generally referred to as a
complete grasp of meaning in a written text in which a dynamic
and growing appreciation of interrelationships in the text is
required (Yang, 2002). In the current views of L2 reading,
much of what the readers do is the same as when they read in
their first language (L1). However, L2 reading could be some-
what slower and less successful than L1 reading, for example,
because of range and size of vocabulary knowledge and cultural
or background knowledge of L2 and the extent of differences
from L1 (A. D. Cohen, 1994; Grabe, 1999). Furthermore, reading
success depends on factors such as the levels of readers’ profi-
ciency, types of text, text difficulty, and task demands (Alderson,
2000).
Lower-level reading includes, for example, automatic word
recognition and automatic recognition of syntactic structures
and parts of speech. Skillful readers can perform certain basic
reading procedures, such as decoding and rereading without
directed attention, with minimal expenditure of cognitive effort.
Some reading processes can be highly automatized, particularly
in familiar environments or after extensive practice. By contrast,
higher-level reading includes aspects of deliberation, goal-
directed attention, and problem solving. For example, L2 readers
occasionally confront an unfamiliar word, syntactic structure, or
topic that requires them to consciously or intentionally evaluate
the sound/symbol relationships, examine alternative sources,
or use context clues. Higher-level reading includes the use of
strategies.
Given the notion of interactions between L2 learners and
cognitive tasks, researchers have attempted to understand the
nature of strategies in L2 contexts (e.g., O’Malley & Chamot,
1990; Oxford, 1990; Purpura, 1999). L2 reading strategy
research has also revealed how strategic readers interact with
a written text and how their strategic behavior is related to
effective reading comprehension (e.g., Anderson, 1991; Barnett,
1988; Bernhardt, 1986; Block, 1986, 1992; Carrell, 1984, 1989,
 Phakiti 651

1991; Salataci & Akyel, 2002). Cognitive and metacognitive strat-

egies are viewed as relevant to L2 reading performance. Cogni-
tive strategies are directly related to the target language and
world knowledge of the learners, which allow them to construct
meaning from text and to perform the given tasks. Cognitive
strategies include making predictions, translating, summariz-
ing, linking with prior knowledge or experience, applying gram-
mar rules, and guessing meaning from contexts. Metacognitive
strategies are related to self-management or self-regulation in a
given reading activity. Metacognitive strategies include plan-
ning and monitoring strategies. Planning strategies refer to
L2 learners’ actions of previewing or overviewing tasks to
develop directions of what needs to be done, and how and when
to do it. Monitoring strategies are learners’ actions undertaken
to check, monitor, and evaluate their thinking and reading
performance.
Recently, researchers have aimed to investigate how male
and female learners differ in L2 reading performance and strat-
egy use (e.g., Bügel & Buunk, 1996; Chavez, 2001; Young &
Oxford, 1997). It is hoped that understanding the ways in
which the strategic reading behavior of males and females influ-
ences L2 reading comprehension will result in the rigorous con-
struction of a sound L2 reading theory. Chavez (2001) points out
that the ultimate goals of understanding gender differences in
strategy use are, for example: (a) to make us aware of how
gender can affect development and achievement in L2 reading;
(b) to enable L2 teachers to use this awareness to help their
students of either gender to achieve gains in L2 reading compre-
hension; (c) to encourage further research into the role of gender
in L2 reading; and (d) to accommodate individual students’
needs, given that males and females deserve an equal chance
of learning success.
To date, there has not been sufficient research that exam-
ines the connection between gender differences in L2 reading
comprehension performance and those in strategy use (as dis-
cussed in the following section). If males and females do indeed
652 Language Learning Vol. 53, No. 4

differ both in their strategy use and L2 reading performance, it

is important to understand the correlation between differences
in strategy use and differences in reading comprehension per-
formance. Furthermore, there is a need for continued research
that can add to our understanding of the specific conditions and
variables that influence males’ and females’ cognitive and meta-
cognitive strategy use and L2 reading comprehension. The pres-
ent study looks at cognitive and metacognitive strategy use by
gender in an official reading test used to make high-stakes
decisions regarding students’ achievement at a Thai university.
The next section discusses some previous research on gender
differences in L2 reading comprehension performance and learn-
ing strategy use. This review follows Chavez (2001), which pro-
vides an extensive discussion of the literature.

Review of the Literature

Previous Research on Gender Differences in L2 Reading

Comprehension

The literature on gender differences in L2 reading perfor-

mance is relatively scarce, and its findings are not consistent. In
some studies, gender differences were not detected. Spurling and
Ilyin (1985) found no gender differences in reading test perfor-
mance among L2 learners. Other studies have found that
females outperformed males. Gardner and Lambert (1972)
found that females performed better than males in French voca-
bulary. Wen and Johnson (1997) found that females demon-
strated higher levels of performance than males on a
standardized national proficiency test. Chavez (2001) found
that regardless of topics and genres, females scored higher
than males in a multiple-choice reading test.
Yet another set of studies found that males perform better
than females. Boyle (1987) found that although female Chinese
university students of English as a foreign language (EFL)
scored higher overall in a language proficiency test, males scored
 Phakiti 653

higher than females on listening vocabulary, which might be

related to reading comprehension. Scarcella and Zimmerman
(1998) found that males performed significantly better than
females on specific types of vocabulary tests. Bügel and Buunk
(1996) found that males also performed significantly better
than females in a reading test that was composed of a gender-
neutral passage. To sum up, research in this area has not
produced conclusive evidence of gender differences in L2 reading
performance, which suggests the need for further research in
the area.

Previous Research on Gender Differences in Strategic Behavior

It should be noted that the nature of research in strategy

use may differ from study to study depending on the specific
language skills being examined. For example, strategy use
research that is conducted in an oral, communicative setting
can differ from that conducted in a silent environment of read-
ing. Most studies investigating gender differences in strategic
behavior are based on the Strategy Inventory of Language
Learning (SILL) approach (Oxford, 1990). Language learning
strategy research focuses on perceived strategy use divorced
from a concrete task (i.e., context-free; Chavez, 2001). Few stud-
ies, however, look at gender differences in actual strategy use in
specific L2 settings. In this article, SILL-based and SILL-related
research on gender differences will be discussed first, followed by
research on actual strategy use in a specific L2 use context.
SILL-based or SILL-related research. Despite some limita-
tions in SILL-based studies on gender (e.g., their tendency to
look at broad patterns of strategy use; see Green & Oxford,
1995), these studies make up the bulk of research on gender
differences in learning strategy use. Oxford and Nyikos (1989),
Oxford, Park-Oh, Ito, and Sumrall (1993), and Oxford, Lavine,
Felkins, Hollaway, and Saleh (1996) found that females reported
more use of cognitive strategies than males. A number of studies
654 Language Learning Vol. 53, No. 4

(e.g., Bacon, 1992a, 1992b; Ehrman & Oxford, 1989; Green, 1991;
Green & Oxford, 1995; Kaylani, 1996; Oxford, 1993; Oxford,
Nyikos, & Ehrman, 1988; Sheorey, 1999) found that females
reported significantly higher use of metacognitive strategies
than males. In brief, a sizable body of SILL-based research has
shown that females report more use of learning strategies than
males (see Chavez, 2001; Young & Oxford, 1997, for detailed
discussion).
Research on actual strategy use in a specific L2 use context. In
contrast to the comparatively plentiful SILL gender-based
research, little research has examined gender differences in
actual strategy use in a specific L2 task. Vandergrift (1997)
employed think-aloud protocols to study gender differences in
listening strategy use among 21 high school learners of French.
He found that there were very few differences when reported
strategy use in the given listening tasks for all male and all
female participants was compared. Females used slightly more
metacognitive strategies than males, whereas males used more
cognitive strategies than females. However, Vandergrift did not
establish a relationship between gender differences in strategy
use and differences in listening performance.
Young and Oxford (1997) investigated strategy use by 23
males and 26 females in processing text in both their L1
(English) and a foreign language (Spanish). Their reading com-
prehension was assessed in open-ended oral recall protocols. The
students were also asked to report their comprehension level
and the use of background knowledge. Young and Oxford found
that males and females did not differ from one another signifi-
cantly in comprehension of the three passages. They also found
no significant overall differences in strategy use between males
and females. However, males reported monitoring their reading
pace, reading strategies, and paraphrasing strategies more often
than females on one L2 passage. Females, by contrast, tried to
solve vocabulary problems and acknowledged their lack of back-
ground knowledge more often than males. Young and Oxford
concluded that gender-based differences in strategic behavior
 Phakiti 655

might not reside in general categories, but rather at the level of

specific strategies. They suggested that some strategies might be
gender-related.
In summary, a review of the literature reveals a void that
future research should be designed to fill. Establishment of a
relationship between strategic behavior by gender and L2 read-
ing comprehension performance is what has been missing in
gender research.

Implications for the Present Study

A number of theoretical and methodological implications

can be drawn from the literature. First, despite the number of
studies that show how strategy use is related to L2 reading
comprehension performance, the connection between gender dif-
ferences in L2 reading performance and those in strategy use
has not been investigated sufficiently or with enough respect to
quality, quantity, and variety in research methodology and focus
(Chavez, 2001). If males and females do indeed tend to use
strategies differently, future research needs to investigate the
relationship of their actual strategy use to their L2 reading
comprehension performance.
Second, considering learners as representative of one or
the other of a pair of dichotomous types (male or female) may
not be sufficient for the development of a gender-based theory,
because aggregation over participants (regardless of achieve-
ment levels) may alter the findings on gender differences in
performance and strategy use. Conclusions based on such
aggregation need to be considered with caution. Hence, the
present study also attempts to look at males and females as
highly successful, moderately successful, or unsuccessful (as
discussed in the Method section).
Third, in the area of methodological approaches, most
previous research analyzed the differences between males and
females in learning strategy use and/or L2 performance by
656 Language Learning Vol. 53, No. 4

means of, for instance, frequency counts, analysis of variance

(ANOVA), or t-tests. Applications of multivariate analysis of
variance (MANOVA), however, seem to better suit an investiga-
tion of gender differences in L2 performance and strategy use
when several dependent factors (e.g., reading comprehension
performance and cognitive and metacognitive strategies) and
independent factors (e.g., achievement levels and gender) need
to be considered simultaneously. The present study aims to
employ this type of statistical analysis.
Finally, as regards contextual issues, there is a lack of
empirical research investigating gender differences in cogni-
tive and metacognitive strategy use under high-stakes read-
ing test conditions. Although Young and Oxford (1997) found
no gender differences in either reading comprehension per-
formance or strategy use, their study was situated in a non-
test, low-stakes reading context. In a high-stakes test
situation, learners’ strategy use may be different because
the test has a gatekeeping function wherein access to certain
achievement grades is restricted (see A. D. Cohen, 1998a,
1998b, for discussion of differences between tests as research
elicitation tools and tests as decision-making devices; see
Shohamy, 2001, for discussion of impacts of tests). In add-
ition, some strategies are specifically used in test-taking con-
texts (e.g., how test takers proceed through the test to
produce answers to questions or test tasks wisely; see
A. D. Cohen, 1998a, 1998b, for discussion of test-taking strat-
egy research). Purpura (1997, 1999) has shown empirically
how cognitive and metacognitive strategies are related to L2
test performance. However, gender differences in strategy
use and L2 performance were not the focus of his studies.
Accordingly, without knowing whether strategy use differs
between test conditions and regular reading conditions and
whether there are gender differences that interact with read-
ing comprehension, the reliability and validity of an L2 read-
ing test cannot be fully asserted. Since each study adds to our
understanding of the specific conditions and variables that
 Phakiti 657

influence males’ and females’ cognitive and metacognitive

strategy use and reading comprehension, continued research
in this area is needed.

Research Questions

The present study aims to answer the following research

questions:
1. Are there statistically significant gender differences in
L2 reading comprehension performance assessed by a
multiple-choice reading comprehension test?
2. Are there statistically significant gender differences in
the use of cognitive and metacognitive strategies for the
completion of such a test?
3. How do males and females at the same achievement
levels differ in reading comprehension performance and in
cognitive and metacognitive strategy use?

Method

Background and Participants

The study was carried out at one of the major government

universities in the north of Thailand using a required basic
English course (Fundamental English I). The data were gathered
during a final examination. The final reading test accounted for
40% of the overall course assessment. Other achievement
assessments included a midterm reading test (35%), listening
comprehension tests (15%), quizzes (5%), and other take-home
assignments (5%).
Five hundred twenty students voluntarily agreed to partici-
pate in this study. The learners took the university’s final
examination in English, immediately followed by a questionnaire
(described in the next section) on their strategy use. However,
658 Language Learning Vol. 53, No. 4

because of incomplete questionnaire data for some test takers

and the elimination of misfitting test takers1 based on test analy-
sis, the number of the participants dropped to 384. The 384
Thai, non–English major, first-year undergraduates were made
up of 173 males (45%) and 211 females (55%). For the purposes
of this study, students who obtained scores of 70% or above on
the examination were classified as highly successful, those who
scored between 45% and 69% were classified as moderately
successful, and those who scored below 45% were classified as
unsuccessful. Among the 384 participants, there were 75 highly
successful (43 females and 32 males), 256 moderately successful
(144 females and 112 males), and 53 unsuccessful (24 females
and 29 males) test takers. The learners were between the ages of
17 and 21 and had been studying English in Thailand for
approximately 8 years. Based on their English test scores from
their university entrance examination, their English proficiency
levels ranged from advanced beginners to intermediate. Students
were given 3 hr to complete the reading test.

Measurement Instruments

Two research instruments were used in the study: a

multiple-choice reading comprehension test and a cognitive-
metacognitive strategy questionnaire.
Assessing reading comprehension. According to Alderson
(2000), different testing techniques for reading may permit the
measurement of different aspects of reading constructs. Based
on Alderson (2000), there is no one best format for reading tests,
and each format has its own strengths and weaknesses.
Although a number of test formats such as cloze tests, gap-filling
tests, multiple-choice tests, constructed-response tests, free-
recall tests, and summary tests have been developed and used
to assess reading comprehension, in the present study, a
multiple-choice test format was selected for use. Despite some
limitations (see Bennett & Ward, 1993; Weir, 1993), the
multiple-choice format is a popular conventional test format for
 Phakiti 659

assessing L2 reading comprehension because it is convenient

to administer and there are well-established procedures for
analyzing multiple-choice test items (Alderson, 2000; Wesche &
Paribakht, 1996). It should be noted that at the university in
which the study was conducted, the authorities decided that the
multiple-choice format was most suitable for assessment purposes.
This test format is thought to have certain advantages over other
test formats including efficiency and inexpensive scoring. It also
allows broad content coverage and control of the range of possible
answers to comprehension questions. To some degree, it controls
test takers’ thinking processes when they are responding to the
test tasks. There are also statistical supports for the analysis of
multiple-choice tests and straightforward interpretations of test
analysis results.
The reading test was composed of two main sections,
namely, a gap-filling cloze and a reading comprehension section
(see Appendix A for the objectives of the basic English course
and information about the test). Prior to the actual test admin-
istration, this test was developed and piloted for content by
the researcher and the teachers of the course, and reliability
analyses were conducted. After the test was implemented in
the main study and scored and the scores were double-checked,
the test data were entered into a computer and analyzed using the
Rasch model of item response theory (IRT; see, e.g., McNamara,
1996, for discussion of Rasch IRT) using the Quest program
(ACER, 1996). The following analyses were performed: internal
consistency (i.e., reliability), item difficulty (i.e., the proportion
of candidates getting an item correct), person ability (i.e., the
relative threshold levels of the test taker’s ability), and item
discrimination analysis (i.e., how well the test’s distracters
work).
The results of the IRT analysis (see Table 1) indicated an
approximate reliability of 0.88 (Part 1: 0.80, Part 2: 0.78). The
correction-for-attenuation correlation (Hatch & Lazaraton, 1991,
p. 444) between parts 1 and 2 was 0.82, which provided some
evidence of construct validity. The relationship found in this
660 Language Learning Vol. 53, No. 4

Table 1

Summary statistics for the reading comprehension test

Parts 1 and 2 Part 1 Part 2

Number of items 85.00 45.00 40.00

Maximum raw score 77.00 41.00 36.00
Mean logit score 00.67 00.68 00.66
SD (logits) 00.74 00.80 00.80
Person separation reliability 00.88 00.80 00.78
(equivalent to KR-20)

study between the scores on the multiple-choice cloze and the

multiple-choice reading comprehension test is similar to that
found by Chapelle and Abraham (1990). The item difficulty and
person ability map of the IRT model indicated a reasonably good
match between the test takers’ ability levels and the test items.
Assessing cognitive and metacognitive strategy use. The
preliminary cognitive and metacognitive taxonomy used in the
present study is presented in Appendix B. The distinction made
in this taxonomy between planning and monitoring strategies is
similar to that employed in O’Malley and Chamot’s (1990) tax-
onomy. Goal setting is viewed as part of planning, whereas
evaluation is considered part of monitoring. From both a meth-
odological and an ethical perspective, strategies used by test
takers in an official test situation may need to be investigated
after test takers have completed the test, rather than during the
test itself, for the following reasons:
1. Requiring the test takers to perform the secondary task
of identifying strategies while taking the test may interfere
with the primary task of reading and completing the test.
The nature of the high-stakes test-taking situation would
thus raise ethical concerns were the participants asked to
report on their strategy use during the test.
2. Relative-frequency (cumulative) strategy use for the
whole test was investigated. Relative-frequency strategy
 Phakiti 661

use implies the degree to which test takers used strategies

overall, rather than the number of times they used a par-
ticular strategy in completing the test. When reflecting on
and assigning strategy use after the fact on the question-
naires employed in this study, test takers are assumed first
to retrieve from memory the degrees of the concurrent
strategy frequencies and second to express them on an
ordinal scale. This method reflects ecological factors per-
taining to the entire test, such as the characteristics of
reading passages, test questions, and difficulty.
3. Having the test takers assess strategy use after the test
allows the researcher to assume that test takers’ overall
reading comprehension performance is directly influenced
by their strategy use in a natural way.
Methods typically used to collect data for use in under-
standing the nature of strategies include verbal reports (e.g.,
think-aloud protocols, retrospective interviews; see A. D.
Cohen, 1996) and self-reported questionnaires (see Oxford,
1996). In the present study, despite some limitations, the choice
of questionnaires to elicit strategy use was considered suitable
because of:
1. the practical and ethical concerns discussed above
2. the difficulty or impossibility of tape-recording all infor-
mants during the test
3. the complexity of verbal-report methodology
4. issues of statistical comparability of learners across groups
Many scholars (e.g., O’Malley & Chamot, 1990; Oxford,
1996; Purpura, 1999) have supported the use of strategy ques-
tionnaires to assess the construct of cognitive and metacognitive
strategies. The questionnaire used in this study allowed learners
to mark strategy use on a 5-point Likert scale: 1 (Never),
2 (Sometimes), 3 (Often), 4 (Usually), and 5 (Always). The length
of time needed to complete the questionnaire ranged from 5 to
662 Language Learning Vol. 53, No. 4

10 min. The following steps were taken to reduce the potential

for invalidity associated with this method:
1. Strategy items used on the questionnaire were drawn
from the literature on L2 reading, learning strategies, and
test-taking strategies, and items were embedded in a model
of human information processing (Gagné, Yekovich, &
Yekovich, 1993).
2. Questionnaire items were presented to participants in
Thai (back translation was incorporated).
3. The questionnaire was piloted with the same population
during their midterm reading test for item-level analysis
(see Phakiti, 2003). Volunteers from among the students
were asked to reflect on how well they understood the
strategy items on the questionnaire.
4. Two experts in metacognition not involved in the present
study were asked to determine to which category each
strategy item included on the questionnaire belonged (i.e.,
cognitive or metacognitive; see Appendix C for the ques-
tionnaire used). Their responses were scored as 1 or 0 in
association with the strategy categories assumed. Experts 1
and 2 had 91.4% and 85.7% agreement with the strategy
categorizations, respectively.
5. An exploratory factor analysis (EFA) was employed to
extract cognitive and metacognitive strategy factors from
the strategy questionnaire data (see Phakiti, 2003, for further
discussion of the EFA validation procedures used in the
present study).
Table 2 presents the composites for cognitive and meta-
cognitive strategies with the estimates of the reliability of the
questionnaire instrument. The approximate overall reliability
estimate for the strategy items was acceptable (0.88) given
the number of the items and the underlying constructs. The
composites of the metacognitive and cognitive strategies
 Phakiti 663

Table 2

Composites of cognitive and metacognitive strategies with reliability

estimates
Number
Strategies of items Items used Reliability

Cognitive 5 2, 6, 7, 8, 9 0.75
Metacognitive 10 14, 15, 19, 21, 22, 24, 30, 31, 32, 33 0.85
Total 15 0.88

presented in the table will be used in the MANOVAs and post

hoc tests.

Data Analyses

A significance level of 0.05 (p < 0.05) was set. Statistical

Package for the Social Sciences (SPSS) version 10 for personal
computers (SPSS Inc., 2001) was used to compute descriptive
statistics and perform reliability analyses, EFA, and MANOVA.
(For further discussion of MANOVA, see Diekhoff, 1992; Stevens,
1996). Following Bollen and Lennox (1991), the data in the pres-
ent study can be considered an emergent variable system in
which the underlying (or emergent) construct (the resultant com-
position of the outcome variables) is dependent on the observed
variables. MANOVA is well suited for use with emergent variable
systems (e.g., Bollen & Lennox, 1991; Cole, Maxwell, Arvey, &
Salas, 1993). Based on the criteria given by J. Cohen (1992), the
sample size in this study is adequate for attaining the desired
statistical power in MANOVA. To answer the first and second
research questions, learners were categorized as males or
females. To answer the third research question, learners were
further classified by achievement levels, with the following groups
resulting: (a) highly successful females, (b) moderately successful
females, (c) unsuccessful females, (d) highly successful males,
(e) moderately successful males, and (f) unsuccessful males.
664 Language Learning Vol. 53, No. 4

Table 3

Descriptive statistics for the item-level cognitive and metacognitive

strategies
Item Mean SD Median Mode Skewness Kurtosis

2 3.71 0.99 4.00 4.00
0.44
0.41

6 3.96 0.93 4.00 4.00
0.56
0.47
7 3.30 1.02 3.00 3.00
0.08
0.46
8 3.56 0.97 4.00 4.00
0.22
0.48
9 3.75 1.02 4.00 4.00
0.45
0.46
14 3.72 0.89 4.00 4.00
0.27
0.17
15 3.37 0.95 3.00 3.00
0.12
0.33
19 3.90 0.93 4.00 4.00
0.49
0.43
21 3.26 0.88 3.00 3.00
0.04
0.15
22 3.21 0.86 3.00 3.00
0.39 0.11
24 3.62 0.91 4.00 4.00
0.12
0.59
30 3.65 0.85 4.00 4.00
0.31 0.12
31 3.58 0.93 4.00 4.00
0.29
0.33
32 3.47 0.88 3.00 3.00
0.18 0.01
33 3.81 0.82 4.00 4.00
0.17
0.61

Results and Discussion

Descriptive Statistics

Table 3 presents descriptive statistics for each of the cogni-

tive and metacognitive strategies at the item level. Table 4
summarizes the descriptive statistics for the test scores and cognitive

Table 4

Descriptive statistics for EFL reading comprehension and cognitive

and metacognitive strategy use variables
Variable Mean SD Median Mode Skewness Kurtosis

EFL reading scores 48.65 10.01 49.00 45.00
0.26 0.30

Cognitive strategies 3.66 0.69 3.60 4.00
0.29
0.44
Metacognitive strategies 3.56 0.58 3.60 3.00 0.04
0.47
 Phakiti 665

and metacognitive strategy use variables. All the variables were

found to be normally distributed (skewness and kurtosis were near
zero). Table 5 presents the estimated marginal means for the reading
scores and the strategy use by achievement and gender.

Gender Differences in Reading Test Performance and Strategy Use

Results related to research questions 1 and 2. To interpret

the univariate F-tests for the different groups, three statistical
tests were performed to assess whether the statistical assumptions
underlying the use of MANOVA were violated in the data set.
First, Box’s M test of equality of covariance matrices demonstrated
that the data had homogeneity of variance (i.e., this test was not
significant at an alpha level of 0.001). Secondly, Levene’s test of
equality of error variances indicated that the homogeneity of
variance for each of the dependent measures was not violated in
the data set (p > 0.001). The third test used was the multivariate
test of significance, Pillai’s trace criterion variance, one of the
most robust statistical tests against violations of assumptions.
The result of this test indicated that there was a statistically
significant multivariate effect for gender levels (F ¼ 3.099,
p < 0.05, Z2 ¼ 0.024).2 Based on the results of these statistical
procedures, the univariate F-tests for the different groups can be
considered statistically significant and therefore meaningful.
Having determined that the results met the statistical cri-
teria set out above, the next step was to check whether there was
a statistical interaction effect between the independent vari-
ables, known as between-participants effects, by examining the
Achievement  Gender source. An interaction effect (when p < 0.05)
implies that if a statistically significant difference between
gender in L2 reading performance and strategy use is found,
both the achievement-level and gender factors are responsible
for the difference. If that is the case, any significant result
dealing with these two variables needs to be interpreted cau-
tiously. In this analysis, there was no interaction effect
(p > 0.05). Hence, statistically significant gender differences
Table 5

Estimated marginal means of reading test performance and strategy use by achievement success and gender
Dependent Success 95% confidence interval
variable levels/Genders Mean Standard error Lower bound Upper bound

Reading performance Highly successful females 62.86 0.77 61.35 64.37

Moderately successful females 48.23 0.42 47.40 49.06
Unsuccessful females 31.42 1.03 29.39 33.44
Highly successful males 62.00 0.89 60.25 63.75
Moderately successful males 47.63 0.47 46.70 48.56
Unsuccessful males 32.83 0.94 30.99 34.67
Total Female 47.58 0.45 46.70 48.45
Male 47.44 0.46 46.53 48.35

Cognitive strategies Highly successful females 3.78 0.10 3.58 3.97

Moderately successful females 3.74 0.06 3.63 3.84
Unsuccessful females 3.05 0.13 2.79 3.31
Highly successful males 4.06 0.12 3.83 4.28
Moderately successful males 3.64 0.06 3.52 3.76
Unsuccessful males 3.22 0.12 2.98 3.46
Total Female 3.54 0.06 3.42 3.65
Male 3.63 0.06 3.51 3.75
Table 5 (continued)

Estimated marginal means of reading test performance and strategy use by achievement success and gender
Dependent Gender success 95% confidence interval
variable levels/Genders Mean Standard error Lower bound Upper bound

Metacognitive strategies Highly successful females 3.76 0.08 3.60 3.92

Moderately successful females 3.53 0.04 3.44 3.62
Unsuccessful females 2.98 0.11 2.77 3.20
Highly successful males 4.10 0.09 3.91 4.28
Moderately successful males 3.58 0.05 3.49 3.68
Unsuccessful males 3.16 0.10 2.97 3.36
Total Female 3.43 0.05 3.33 3.52
Male 3.62 0.05 3.52 3.71
668 Language Learning Vol. 53, No. 4

found in this analysis are due to the main effect (i.e., gender)
only.
Table 6 presents the results of the MANOVA for gender
differences in the reading test performance and cognitive and
metacognitive strategy use. Metacognitive strategy use was the
only variable that differed significantly between genders. Based
on pairwise comparisons using estimated marginal means,
males reported significantly higher use of metacognitive strat-
egies than females.
Therefore, in response to the first research question (Are
there statistically significant gender differences in L2 reading
comprehension performance assessed by a multiple-choice read-
ing comprehension test?), it was found that males and females
did not differ in their reading comprehension performance. This
finding is similar to those of Spurling and Ilyin (1985) but dif-
ferent from those in other studies in the literature (e.g., Boyle,
1987; Bügel & Buunk, 1996; Chavez, 2001; Gardner & Lambert,
1972; Scarcella & Zimmerman, 1998; Wen & Johnson, 1997). In
response to the second research question (Are there statistically
significant gender differences in the use of cognitive and meta-
cognitive strategies for the completion of such a test?), males and
females were not found to differ in their cognitive strategy use.
This finding differs from those of most SILL-based studies dis-
cussed in the literature review. Unexpectedly, however, males
were found to report significantly more use of metacognitive
strategies than females. This finding is again not consistent
with those in most previous learning strategy research.

Table 6

Factorial MANOVA results (1)

Source Dependent variables df F p Z2

Gender EFL reading performance 1 0.05 0.855 0.000

Cognitive strategies 1 1.16 0.192 0.003
Metacognitive strategies 1 8.02 0.005 0.021
 Phakiti 669

Note that the L2 learners in this study may be different

from those in previous studies. For example, they were all from
the same age group, had a relatively small range of English
proficiency levels, and shared the same cultural and educational
context (i.e., the group exhibited homogeneity). Unlike partici-
pants in much research in English L2 reading and learning
strategy use, these learners had a nonalphabetic L1 coding sys-
tem. Although caution is needed when comparing research find-
ings across studies, in regard to L2 reading performance, the
present findings are similar to those of Young and Oxford (1997),
who look at reading under nontest L2 reading conditions, in that
there are no gender differences in L2 reading comprehension
performance. However, in the present study, males were found
to report significantly more use of metacognitive strategies than
females (further discussed in the following section). Note that L1
backgrounds differ in this study when compared to those of
Young and Oxford’s participants.
There are three possible interpretations for the present
findings on gender differences in metacognitive strategy use.
First, in the gender-based literature, females are seen as better
language learners than males in L2 learning (see Chavez, 2001).
The higher use of metacognitive strategies by men than women
in the current study might explain why females did not outper-
form males on the test, or why males attained the same levels of
reading performance as females. Phakiti (2003) found that the
use of cognitive and metacognitive strategies among these stu-
dents explained 15% to 22% of the variance in reading compre-
hension scores. Based on Phakiti (2003), the degree of the
relationship between strategy use and reading comprehension
performance yielded a meaningful interpretation of the relation-
ship of strategy to reading comprehension performance.
Second, given that learners are likely to be aware of their
performance when they are faced with difficulty and considering
females’ equal reading comprehension performance in the
current study, females might not have encountered as much
difficulty during the reading test as males did. If this were the
670 Language Learning Vol. 53, No. 4

case, females’ metacognitive strategy use would play a limited

role. Therefore, they might report less use of metacognitive
strategies. To my knowledge, an interpretation from this angle
in gender-related strategy research has rarely been discussed.
According to Alexander, Graham, and Harris (1998), when learn-
ers move from acclimation to competence to expertise, there is a
shift in the type of strategies they need and use. In the course of
strategy development, as learners gradually and continuously
increase their L2 competence and become more capable at L2
reading, the strategies they use and the way they use them
change. While they are new to the L2 and as they are beginning
to develop their reading competence, their strategy use in gen-
eral may be highly conscious but may often be inefficient and
unsophisticated because mastery of the L2 remains incomplete
and fragmented. At this conscious incompetence stage, although
strategies can be highly conscious, they do not necessarily guar-
antee successful reading performance. Perhaps this explains
why less successful learners choose less effective strategies and
use them less effectively than more successful learners. When
they move from a level of nonmastery to mastery and some
specific strategies demand less cognitive energy or conscious
processing, their strategic processing becomes skillful processing
(i.e., unconscious competence).
Third, although anxiety is not a focus of the present study,
it is worth noting that the bulk of psychological research has
pointed to the conclusion that females have higher test anxiety
than males (see Zeidner, 1998). Some empirical data have also
supported the hypothesis that males and females interpret and
respond to a test situation in different manners (e.g., Couch,
Garber, & Turner, 1983; Lewis & College, 1987). Males may be
more likely than females to perceive a test situation as a personal
challenge. Females, by contrast, may perceive the test situation
as a threat, leading to states of fear, worry, and low self-esteem.
If these interpretations are correct, the present finding might
have three useful implications: First, teachers could help accom-
modate female students’ needs by means of metacognitive training
 Phakiti 671

in test conditions; second, language testers may need to further

consider concurrent issues of fairness and equity between
males and females (e.g., Willingham & Cole, 1997) and to
design tests that incorporate an awareness of strategic
capabilities; and third, if females tend to be more metacognitive
than males in a nontest reading situation and males tend to be
more metacognitive than females in a reading test situation,
then the nature of the test context might suggest the use of
somewhat different kinds of metacognitive strategies by the two
genders. If they differ greatly, to what extent can findings in
metacognitive strategy use by gender from reading in high-stakes
test contexts be transferred to that in low-stakes reading set-
tings? This question merits further attention in future research
regarding test validity (Messick, 1989). These three possible
interpretations, however, require further empirical validation.
Nonetheless, viewing learners according to their dichot-
omous gender type may not be sufficient for development of a
gender-based theory, because learners are not just males or
females, but they are also successful or unsuccessful on the
test. The next section addresses this concern.
Results related to research question 3. In response to the
third research question (How do males and females at the same
achievement levels differ in reading comprehension performance
and in cognitive and metacognitive strategy use?), MANOVA
was used to examine gender differences across achievement
levels. Box’s M test of equality of covariance matrices demon-
strated that the data had homogeneity of variance. Levene’s test
of equality of error variances indicated that the homogeneity of
variance for the dependent measures was not violated in the
data set. The multivariate tests of significance, Pillai’s trace
criterion variance, indicated that there was a statistically sig-
nificant multivariate effect for gender levels (F ¼ 27.732,
p < 0.05, Z2 ¼ 0.268). Table 7 presents the results of the
MANOVA, which suggested statistically significant differences
in EFL reading performance and cognitive and metacognitive
strategy use across the six groups.
672 Language Learning Vol. 53, No. 4

Table 7

Factorial MANOVA results (2)

Source Dependent variables df F p Z2

Gender across EFL reading test performance 5 227.37 0.000 0.750

success levels Cognitive strategies 5 9.75 0.000 0.114
Metacognitive strategies 5 16.74 0.000 0.181

Scheffé post hoc tests were therefore conducted to point

out statistically significant differences among the dependent
variables. Males and females across Gender  Achievement groups
showed both similarities and differences in their reading perfor-
mance and cognitive and metacognitive strategy use. For example,
highly successful males and females reported significantly higher
use of cognitive and metacognitive strategies than unsuccessful
males and females. For the purpose of this article, Table 8 presents
only the results for post hoc tests that compare genders at the
same achievement level. As can be seen in Table 8, test takers at
the same achievement levels across gender did not differ either
in their reading performance or use of cognitive and meta-
cognitive strategies. All the findings above provide detailed
empirical justifications for an evaluation of the degree of validity
and fairness of particular inferences derived from the test.
Perhaps the findings here are useful in suggesting that aggrega-
tion over participants may have a potential to alter the findings
on gender differences in L2 performance and strategy use, as
found in answers related to research questions 1 and 2.
The findings of this study confirm those made by O’Malley
and Chamot (1990) in that the highly successful learners
reported significantly higher use of metacognitive strategies
than the unsuccessful learners. However, O’Malley and Chamot
(1990) found no significant difference in cognitive strategy use
between successful and unsuccessful learners. The study pre-
sented here has also provided empirical support for Bachman
and Palmer’s (1996) model of language ability in that the use
Table 8

Scheffé post hoc test of gender differences between like achievement groups
Dependent variable Gender success (I) Gender success (J) I–J Standard error p

EFL reading Highly successful females Highly successful males 0.861 1.18 0.991
performance Moderately successful females Moderately successful males 0.603 0.63 0.970
Unsuccessful females Unsuccessful males
1.411 1.39 0.960

Cognitive strategies Highly successful females Highly successful males
0.280 0.15 0.650
Moderately successful females Moderately successful males 0.099 0.08 0.922
Unsuccessful females Unsuccessful males
0.171 0.18 0.971

Metacognitive Highly successful females Highly successful males
0.339 0.12 0.192
strategies Moderately successful females Moderately successful males
0.053 0.07 0.987
Unsuccessful females Unsuccessful males
0.179 0.15 0.915
674 Language Learning Vol. 53, No. 4

of cognitive and metacognitive strategies helps explain why

successful and unsuccessful learners across gender perform as
they do. The achievement variable seemed mainly responsible
for the statistically significant differences in the reading com-
prehension performance and strategy use found in the analysis
of the results in this study. Further research may need to include
other independent variables in MANOVAs, such as L2 reading
proficiency. Any interaction effect found from the analyses may
yield useful information to explain gender differences.
Note that the distinction between cognitive and metacogni-
tive strategies is more complex than reported in previous strat-
egy research (e.g., O’Malley & Chamot, 1990; Oxford, 1990;
Purpura, 1999). Phakiti (2003) suggests that the goals or
functions of strategy use may determine whether cognitive strat-
egies remain at a cognitive level (e.g., to gain knowledge or to
comprehend texts) or progress to a metacognitive level (e.g., to
set future goals or to monitor comprehension). For example, if
learners translate a passage to memorize the basic gist, transla-
tion at this level will be a cognitive strategy, as asserted by most
reading and strategy researchers. However, if learners inten-
tionally translate the passage because they want to make sure
that they have accurately understood the content before proceed-
ing, translation may be a metacognitive strategy.
Perhaps the difference found in this study between the highly
successful and the unsuccessful readers in the use of cognitive
strategies is that highly successful learners tended to use cognitive
strategies at a metacognitive level, whereas unsuccessful learners
continued to use cognitive strategies at a cognitive level. This
possibility warrants further research, because the ability to use
cognitive strategies as metacognitive strategies may reflect
metastrategic competence (e.g., Phakiti, 2003). Metastrategic
competence implies an understanding of the need to use a strategy
and/or an awareness of the pitfalls of using a strategy in a less
than adequate way. Cognitive and metacognitive strategy use at
the metacognitive level may hence be indistinguishable from one
another. The problem inherent in the cognitive-metacognitive
 Phakiti 675

distinction is apparent when we attempt to apply either of the

terms to a specific instance (Weinert, 1987; Wilson, 2000).
For a number of learners, some strategies that were
formerly cognitive for those learners may no longer function at a
cognitive level, but at a metacognitive one (however, researchers
still view these strategies as cognitive, despite the possibility
that they might be functioning at a metacognitive level in a
given context). This raises the question of whether cognitive
strategies should be viewed as static or dynamic. If they are
dynamic and the level at which they operate changes according
to the current demand, researchers will not always be able to
identify cognitive strategies. Since the problem of identifying
strategies seems to depend on underlying goals or functions
of strategy use, a combination of questionnaire data and
qualitative data (e.g., verbal reports) in future research should
provide useful additional information for fine-tuning the criteria
used to distinguish between cognitive and metacognitive
strategies.

Conclusions and Implications for Future Research

The present study investigated gender differences in L2

reading performance and cognitive and metacognitive strategy
use in an EFL reading test context. It has provided empirical
evidence that although gender differences do not play a signifi-
cant role in reading comprehension performance, they may
nevertheless play a role in cognitive and metacognitive strategy
use.

Limitations

A few limitations of the present study are worth addressing

because they provide agendas for future research. First, the
taxonomy of cognitive and metacognitive strategies employed
in the study included only a partial set of the possible strategies
students might have used during the reading test. Second, the
676 Language Learning Vol. 53, No. 4

scope of investigation was limited. This study did not look at

gender differences in cognitive and metacognitive strategy use
as an item-level strategy (e.g., translation), as Young and Oxford
(1997) suggest may be relevant. Perhaps there may be specific
gender-related strategies detectable by MANOVAs. Third, this
study did not examine the frequency of cognitive and metacog-
nitive strategy use (e.g., high, medium, low) by gender. Fourth,
the study employed a cross-sectional assessment of learners’
strategy use. When a single performance is examined, gender
differences may be more likely to be detected in an inconsistent
way.
Finally, the present study employed only ordinal-scale
questionnaires to assess strategy use. Despite the perspective
that the nature of strategies should be related to the notion of
intentional and purposeful behavior (see Alexander et al., 1998;
Kintsch, 1998; Phakiti, 2003), there is a possibility that the
questionnaire captured automatized or routinized processing,
because when the participants answered the questionnaire,
they reconstructed what actually happened while they were
completing the reading test tasks (post self-analysis). For example,
some learners might have developed an automatic reading
behavior of underlining important ideas in texts. At the time
they were completing the test, they might not have been aware
of this behavior. However, when they answered the question-
naire (here strategy items provided specific prompts about their
thinking), they might have realized that they always under-
lined important ideas in texts and therefore answered always.
Although automatized processing may be relevant to specific
language performance, it is not part of the strategic or meta-
cognitive construct.
Based on Gagné et al. (1993), in contrast to consciously
deployed control processes, which involve goal setting, planning,
monitoring goal attainment, and revising plans, some control
processes (either innately or through overlearning) are deployed
automatically. These processes are not referred to as metacogni-
tive because they do not appear to be beyond the processing
 Phakiti 677

event, but rather, they are part of it. The flow of control in
learned, automated skills is embedded in the skill, although
the sequence has a control structure. Hence, it can be argued
that the effect of automatization may need to be minimized in
the course of strategy use assessment and research.
Future strategy items may need to include some adverbs
such as consciously, intentionally, and deliberately. This none-
theless will not necessarily guarantee that the problem of the
conscious-unconscious distinction will be resolved. Post self-
analysis may always alter findings concerning the nature of
cognitive and metacognitive strategy use. Deriving data from a
research instrument of this sort when post self-analysis is
present could reduce the validity of the strategy definition and
hence of the investigation. Of course, it can be argued that in
the present study, preliminary analyses such as univariate
assumptions and EFAs (i.e., a rigorously determined set of
observed variables that defined the constructs) were conducted
prior to the data use. In a high-stakes test context, access to
learners’ online processing is difficult (see the Method section).
Even if we could employ verbal reports in this context to resolve
the reactive effect, the issue of conscious and unconscious
processing would remain problematic.
Given the nature of the cognitive and metacognitive con-
structs involved in L2 use, the number of possible interactions
among strategies in this operational setting and the data-
gathering methods and analyses, the present study does not
claim to provide a comprehensive look at gender differences in
L2 reading comprehension performance and cognitive and meta-
cognitive strategy use. The present study does, however, have
merit for its theoretical and methodological considerations. The
next section discusses first, the construct of gender, second,
underlying perspectives on the gender construct, and finally,
the nature of strategies. Although conclusive answers regarding
the nature of gender differences in L2 performance and strategy
use cannot be provided, it is hoped that new avenues for future
gender research will be opened.
678 Language Learning Vol. 53, No. 4

Implications for Theory Development

Concept of gender. The discussion of gender as a variable

influencing strategy use often forces us to revisit how gender has
been defined and how interpretations of research findings
regarding gender have been made. In all studies brought
together here, the category gender has been based on a biological
dichotomy. However, when researchers interpret the results of
their studies, the meaning of gender deviates from the biological
differences between males and females to the psychological and
social differences associated with being feminine or masculine.
In other words, male and female identity is partly defined by a
given culture through beliefs and expectations about what it
means to be masculine and feminine. This changing definition
can be related to the concept of gender identity.
The concept of gender identity poses theoretical and prac-
tical problems for strategy research. First, there is no theoretical
approach currently available in the area of L2 performance and
strategy use that can deal adequately with the dynamic nature
of gender identity as it emerges in context. Thus, the static
categories of male/female postulated in the usual biological
dichotomies cannot deal with the range of personal alignments
that are found in actual practice, as individuals position
themselves as being masculine or feminine. This fluidity of
identity makes it difficult for researchers to design instruments
that can capture the strategic nature of gender production and
interpretation.
The shift in the definition of gender may be partly related to
the adaptation of the term from its original educational purpose
(e.g., gender has traditionally been used in schools to segregate
classes, either overtly or in subtle ways) to a more sociopolitical
one (e.g., gender equity; Willingham & Cole, 1997). Once the
purpose of the definition changes, gender is no longer truly
binary. As a result, gender becomes an untidy construct that
can be manipulated statistically and adjusted for sociopolitical
aims. Hence, gender researchers need to question whether
 Phakiti 679

gender as a dichotomous variable is really the concept they have

long pursued. Researchers have been faced with difficulty in
generalizing gender differences across contexts and cultures.
When examining gender differences, researchers can only distill
possible relevant context-related factors to explain gender differ-
ences. The cause of this difficulty seems to be a discrepancy
between a basic universalist, essentialist stance that makes
gender an innate, immutable category and a sort of social-
constructivist framework that sees gender not as innate, but as
socially and cognitively created and maintained.
To resolve this problem and establish a principled agenda
for gender research, I propose the following principles: First, the
static biological dichotomy should be retained when learner
characteristics are being represented. Second, both genders
should be viewed according to their ability/achievement levels.
Third, gender differences should be interpreted very strictly in
their specific contexts. Finally, the belief in gender differences as
a universal phenomenon should be disregarded (as discussed in
the following section). It can be argued that even though it is
possible to establish gender differences as a universal phenom-
enon in L2 performance and strategy use, which would allow
researchers to compare research findings across contexts, what
would be discovered would be whether gender differences exist.
What is more important is where and how such differences
originate.
Theoretical perspectives on gender. Researchers’ perspec-
tives on gender have often gone unanalyzed in the literature.
Males’ and females’ L2 reading comprehension performance in
this study can be interpreted from at least two different perspec-
tives, namely, those of the traditional trait theorist and of the
behaviorist. For the trait theorist, it is not possible that learners
could perform successfully on the test if they did not have the
reading skills required for the level of material the test pre-
sented or did not know how to use certain strategies. In fact,
many gender-based researchers have a tendency to be trait
theorists. This is evident from an implicit underlying principle
680 Language Learning Vol. 53, No. 4

that shows up in their focus on context-free strategy use and

universal gender differences. Nevertheless, the behaviorist who
believes in the influence of contextual factors on performance
would argue that if there are such things as stable L2 reading
competence, strategic ability, and universal gender differences,
why do the findings in the present study suggest that males and
females did not differ in their reading performance, and why did
males report more use of metacognitive strategies than females?
There is evidence to suggest that context in this study influenced
males’ and females’ performance and strategy use. For example,
the test was a timed, high-stakes test. Surely completing a read-
ing test question in a high-stakes test setting is different from
completing the same reading test question in a quiet and sunny
garden. Therefore, the behaviorist could argue that charac-
teristics of the context could potentially determine what
opportunities were lost for learner participation in the high-
stakes activity.
Although the debate between the traditional emphasis on
the individual’s traits and the more radical conception of con-
texts as affecting the distribution of cognition has not yet been
resolved, trait competence (e.g., linguistic and/or strategic) can-
not be thought of without considering concurrent cognitive activ-
ities, whereas state performance is not conceivable without
considering learners’ potential cognitive residues. It can be
argued that these two extreme perspectives need to be accom-
modated within the same theoretical framework in order to
further develop the theory of gender in L2 contexts. Without
this accommodation, no theory of gender is reasonable. To
accommodate these two traditions in the understanding of
gender differences reflects the interactionalist perspective (see
Chapelle, 1998, for further discussion of this perspective).
Within this perspective, the major concern is for whom, under
what conditions, and within what situation a particular strategy
has adaptive consequences on performance. Given this perspec-
tive, perhaps, gender is a state construct because it is changing
and unstable.
 Phakiti 681

Nature of strategies. Strategic competence has been under-

stood as a higher-level cognitive trait that involves active moni-
toring and consequent regulation and orchestration of cognitive
processes to achieve cognitive goals (see Bachman & Palmer,
1996).3 Situations in which strategic competence plays a crucial
role include cognitively demanding (e.g., when a lot of specific
language-related knowledge is required) and high-stakes (e.g.,
affecting the performance of individuals’ lives) situations.
Within these situations, the learners’ deliberate reactions in
response to the difficulty are called strategies. In other words,
strategies are not individuals’ cognitive traits, but specific
actions or behaviors resulting from the regulation of their cogni-
tion. The success of their cognitive regulation also depends on
their language ability (Purpura, 1999). That is, although learn-
ers may be aware of the danger inherent in or difficulty of a
particular problem, if they lack the essential cognitive resources
to resolve it, it is unlikely that they can overcome the danger or
difficulty.
To date, there is little empirical evidence to show how
language learning strategies are related to actual strategy use
in context. To what extent are findings concerning gender differ-
ences in learning strategy use related to findings concerning
actual strategy use in context? As Barnett (1988) points out,
the interaction between strategy use and perceived strategy
use among L2 readers is complex. Barnett found that (a) L2
reading comprehension increased with better use of strategies
in a specific reading context, and (b) comprehension increased as
students perceived that they used more effective strategies, with
or without an emphasis on the strategy of reading through con-
text. Nevertheless, exactly how actual strategy use in context
interacts with perceived strategy use needs further research
(Barnett, 1988).
It can be argued that language learning strategies are not
actually strategies in the strictest sense of the term. Rather,
they should be seen as learners’ stable long-term knowledge
of their strategy use. Perhaps the use of the term language
682 Language Learning Vol. 53, No. 4

learning strategies suggests a misleading consensus on what

the construct actually is. Based on the theory of metacognition
(see Schraw & Moshman, 1995, for further discussion), lan-
guage learning strategies are related to knowledge of cognition
(the accumulated autobiographical information about one’s
own cognition), whereas use strategies are related to regula-
tion of cognition (the ongoing monitoring and regulation of
cognitions). In other words, language learning strategies tend
to be a type of relative-frequency knowledge of strategy use
stored in long-term memory, whereas use strategies are relative-
frequency knowledge related to working memory regarding
concurrent cognitive tasks. Thus, although language learning
strategies and use strategies are likely to be related to one
another, they pertain to different kinds of cognitive-reference
classes. Here, in line with Kintsch (1998), just because learners
know something about their strategy use does not guarantee
that this knowledge is activated in a given cognitive process at
a given time, although it would be relevant for that process. That
is, a sequence of actions that learners know will achieve a
certain cognitive goal needs to be distinguished from a sequence
of actions they actually undertake in a specific context. Since
situational contexts can potentially alter the impact of the
strategy disposition on actual reading performance, to explain
specific L2 performance, actual strategy use will be more
relevant than perceived strategy use.
Given this confusion in terminology, future research needs
to explore the extent to which language learning strategies and
actual use strategies are related. Logically, if there is strategic
behavior demanded by a situation and if learners regularly
exhibit it, then their behavior tells us more about the task
environment than about the learners. If, however, an analysis
of strategic behavior shows patterns of similarity or difference
between the extent to which the typical learners perceive using
certain strategies over a variety of contexts and the extent to
which they actually use them in a specific context, then this pat-
tern represents something about the psychology of strategy use.
 Phakiti 683

Note also that in the case of L2 contexts, because learners’

interlanguage develops at varying rates and exposure to lan-
guage use can differ significantly from person to person even
within the same cultural context, interlanguage and strategies
may need to be perceived as state constructs. When they are
understood as being state, low stability in L2 performance or
strategy use, low test-retest reliabilities of some strategy or
test measures, and inconsistent findings across studies will be
more comprehensible, thereby offering researchers more room to
construct a sound L2 theory.

Conclusions

Previous research in learner strategies by gender has not

taken seriously one or more of the following: the state-trait
strategy distinction, aspects of self-consciousness, goals or
motivation of using strategies, and the role of context. These
aspects are intertwined. First, use strategies are transitory and
situation-specific. Perhaps observations of consistency in use
strategies across tasks and occasions can be used to infer learn-
ers’ strategic traits. Second, strategies differ from skills or other
common processes in that they are related to self-consciousness.
Self-consciousness in strategies, skills, and performance has
a developmental status ranging from conscious incompetence
to conscious competence to unconscious competence. Whereas
strategies seem to exist within the first two developmental
stages, skills are present only within the unconscious competence
stage. Assuming that learners have strategic competence, if
there is no difficulty or problem in completing an L2 task, their
cognitive processing is not strategic, but skillful. The issue of
strategic and skillful reading is, however, complex in the context
of formal L2 learning because learners also need to respond
to given reading tasks such as comprehension questions. The
presentation of reading tasks is likely to activate an awareness
of learners’ reading and cognitive processing. With the goal of
completing the reading tasks in mind, L2 reading becomes
684 Language Learning Vol. 53, No. 4

purposeful and goal-oriented. Given that learners’ reading pro-

cesses can engage both skillful and strategic reading, it is not
clear when reading behaviors are skillful or strategic.
Third, perhaps the real goal of strategy research should be
not to differentiate cognitive strategies from metacognitive strat-
egies, but to understand the underlying goals (cognitive or
metacognitive) or motivations for using a particular strategy.
Strategic behaviors seem to be dynamic. For example, checking
for accuracy (a metacognitive strategy) can take the form of
cognitive strategy use such as readers’ translating a difficult
sentence several times because they feel that the sentence
contains the gist of the whole text. If they understand it, they
can be confident about the correctness of the test answer they
choose. This is an area in which future investigation should
occur. Finally, the concept of ecology (Piaget, 1971) needs to be
highlighted in strategy research. It is beyond the scope of this
article to discuss how the notion of context is often unanalyzed
and left implicit by most researchers. However, investigating
strategy use may provide a key to a better understanding of
context given that its meaning ranges from physical to mental
phenomena (see Butterworth, 1993; Mercer, 1993).
To conclude, since research must begin to accommodate an
understanding of why and under what conditions gender differ-
ences occur, use of a multiple-task, multiple-occasion assessment
method in future gender research may allow a more rigorous
generalization of gender differences in strategy use. The problem
with such an assessment method, however, is related to practi-
cality. That is, it can be expensive and time consuming. Undoubt-
edly, use of such a method will throw some light on a sound
gender theory of strategic behavior. Future studies may wish to
replicate the present study in terms of theoretical and methodo-
logical approaches. The aim of replications should not be primar-
ily to find out whether the findings of the current study would be
the same or different in other contexts, but to understand where
and how gender differences originate. Given that our understand-
ing of gender and strategy use research has been driven by an
 Phakiti 685

increasingly precise analysis of the widening range of strategic

phenomena, it is hoped that the present study can make some
contribution.
Revised version accepted 2 May 2003

Notes
1
Misfitting test takers is a statistical term from Rasch’s item response
theory. Here, fit statistics are expressed as mean square or t-statistics.
Based on McNamara (1996), these statistics enable researchers to
investigate the coherence of a test taker’s responses as part of a set of
responses from a larger group of test takers. Misfitting test takers are
hence those whose abilities are not measured appropriately by this
particular test. In other words, the direction of misfit is of the test to the
test taker, not the test taker to the test.
2
J. Cohen (1977) characterizes eta-squared or the ratio of the
explained variance to total variance as follows: Z2 ¼ 0.01 as small effect
size, Z2 ¼ 0.06 as medium effect size, and Z2 ¼ 0.14 as large effect size.
Medium effect size represents an effect likely to be visible to the naked
eye of a careful observer, whereas small effect size is set to be noticeably
smaller than medium effect size but not so small as to be trivial (J. Cohen,
1992). Large effect size is set to be the same distance above medium as
small is below it. Use of magnitude-of-effect estimates according to Cohen’s
characterization can assist the researcher in establishing whether
statistically significant findings are of practical or meaningful significance
within the context of an empirical investigation, that is, whether the
findings are likely to be a result of an artifact of sample size in the course
of rejecting the null hypothesis (H0).
3
Bachman and Palmer’s (1996) conceptualization of metacognitive
strategies as a way of elaborating the notion of strategic competence
differs from the notion of metacognitive strategies I have discussed in this
article, since what they refer to is the metacognitive function of human
cognition (L. Bachman, personal communication, January 24, 2003).

References

Alderson, J. C. (1979). The cloze procedure and proficiency in English as a

foreign language. TESOL Quarterly, 13, 219–227.
Alderson, J. C. (2000). Assessing reading. Cambridge, England: Cambridge
University Press.
686 Language Learning Vol. 53, No. 4

Alexander, P. A., Graham, S., & Harris, K. R. (1998). A perspective on

strategy research: Progress and prospects. Educational Psychology
Review, 10, 129–153.
Anderson, N. J. (1991). Individual differences in strategy use in second
language reading and testing. Modern Language Journal, 75, 460–472.
Australian Council for Educational Research (ACER). (1996). Quest
Programme [Computer software]. Melbourne, Australia: ACER.
Bachman, L. F., & Palmer, A. S. (1996). Language testing in practice.
Oxford, England: Oxford University Press.
Bacon, S. M. (1992a). The relationship between gender, comprehension,
processing strategies, and cognitive and affective response in foreign
language listening. Modern Language Journal, 76, 160–178.
Bacon, S. M. (1992b). Phrases of listening to authentic input in Spanish: A
descriptive study. Foreign Language Annals, 25, 317–334.
Baker, L., & Brown, A. L. (1984). Metacognitive skills and reading. In
P. D. Pearson (Ed.), Handbook of reading research (pp. 353–394). New
York: Longman.
Barnett, M. A. (1988). Reading through context: How real and perceived
strategy use affects L2 comprehension. Modern Language Journal, 72,
150–162.
Bennett, R., & Ward, W. (Eds.). (1993). Construction versus choice in
cognitive measurement: Issues in constructed response, performance test-
ing, and portfolio assessment. Hillsdale, NJ: Erlbaum.
Bernhardt, E. (1986). Three approaches to reading comprehension in inter-
mediate German. Modern Language Journal, 67, 111–115.
Block, E. L. (1986). The comprehension strategies of second language read-
ers. TESOL Quarterly, 20, 463–493.
Block, E. L. (1992). See how they read: Comprehension monitoring of L1
and L2 readers. TESOL Quarterly, 26, 319–343.
Bollen, K., & Lennox, R. (1991). Conventional wisdom on measurement: A
structural equation perspective. Psychological Bulletin, 110, 305–314.
Boyle, J. P. (1987). Sex differences in listening vocabulary. Language
Learning, 37, 273–284.
Bügel, K., & Buunk, B. P. (1996). Sex differences in foreign language text
comprehension: The role of interests and prior knowledge. Modern Lan-
guage Journal, 80(1), 15–31.
Butterworth, G. (1993). Context and cognition in models of cognitive
growth. In P. Light & G. Butterworth (Eds.), Context and cognition:
Ways of learning and knowing (pp. 1–13). Hillsdale, NJ: Erlbaum.
Carrell, P. L. (1984). Personality and individual writing processes. College
Composition and Communications, 35, 285–300.
 Phakiti 687

Carrell, P. L. (1989). Metacognitive awareness and second language read-

ing. Modern Language Journal, 73, 121–149.
Carrell, P. L. (1991). Second language reading: Reading ability or language
proficiency. Applied Linguistics, 12, 159–179.
Chapelle, C. A. (1998). Construct definition and validity inquiry in SLA
research. In L. F. Bachman & A. D. Cohen (Eds.), Interfaces between
second language acquisition and language testing research (pp. 32–70).
Cambridge, England: Cambridge University Press.
Chapelle, C. A., & Abraham, R. G. (1990). Cloze method: What difference
does it make? Language Testing, 7, 121–146.
Chavez, M. (2001). Gender in the language classroom. Boston: McGraw Hill.
Cohen, A. D. (1994). Assessing language ability in the classroom (2nd ed.).
Boston: Heinle and Heinle.
Cohen, A. D. (1996). Verbal reports as a source of insights into second
language learner strategies. Applied Language Learning, 7, 5–24.
Cohen, A. D. (1998a). Strategies and processes in test taking and SLA. In
L. F. Bachman & A. D. Cohen (Eds.), Interfaces between second language
acquisition and language testing research (pp. 90–111). Cambridge,
England: Cambridge University Press.
Cohen, A. D. (1998b). Strategies in learning and using a second language.
London and New York: Longman.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New
York: Academic Press.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
Cole, D. A., Maxwell, S. E., Arvey, R., & Salas, E. (1993). Multivariate
group comparisons of variable systems: MANOVA and structural
equation modeling. Psychological Bulletin, 114, 174–184.
Couch, J., Garber, T. B., & Turner, W. E. (1983). Facilitating and debilitat-
ing test anxiety and academic achievement. Psychological Reports, 33,
237–244.
Diekhoff, G. (1992). Statistics for the social and behavioral sciences:
Univariate, bivariate, multivariate. Dubuque, IA: Brown.
Ehrman, M., & Oxford, R. (1989). Effects of sex differences, career choice,
and psychological type on adult language learning strategies. Modern
Language Journal, 73, 1–13.
Gagné, E. D., Yekovich, C. W., & Yekovich, F. R. (1993). The cognitive
psychology of schooling learning. New York: HarperCollins College.
Gardner, R. C., & Lambert, W. E. (1972). Attitudes and motivation in
second language learning. Rowley, MA: Newbury House.
Grabe, W. (1999). Developments in reading research and their implications
for computer-adaptive reading assessment. In M. Chalhoub-Deville
688 Language Learning Vol. 53, No. 4

(Ed.), Issues in computer-adaptive testing of reading proficiency

(pp. 11–47). Cambridge, England: Cambridge University Press.
Green, J. M. (1991). Language learning strategies of Puerto Rican univer-
sity students. Paper presented at the annual meeting of Puerto Rico
TESOL, San Juan.
Green, J. M., & Oxford, R. (1995). A closer look at learning strategies, L2
proficiency, and gender. TESOL Quarterly, 29, 261–297.
Hatch, E., & Lazaraton, A. (1991). The research manual: Design and stat-
istics for applied linguistics. Boston: Heinle & Heinle.
Kaylani, C. (1996). The influence of gender and motivation on EFL learning
strategy use in Jordan. In R. Oxford (Ed.), Language learning strategies
around the world: Cross-cultural perspectives (pp. 75–88). Honolulu:
University of Hawai‘i, Second Language Teaching and Curriculum Center.
Kintsch, W. (1998). Comprehension: A paradigm for cognition. Cambridge,
England: Cambridge University Press.
Lewis, E. C., & College, C. (1987). Differential responses of females and
males to evaluative stress: Anxiety, self-esteem, efficacy, and willingness
to participate. In R. Schwarzer, H. M. van der Ploeg, & C. D. Spielberger
(Eds.), Advances in test anxiety research (Vol. 5, pp. 97–106). Lisse, the
Netherlands: Swets & Zeitlinger.
McNamara, T. F. (1996). Measuring second language performance. London
and New York: Longman.
Mercer, N. (1993). Culture, context and the construction of knowledge in the
classroom. In P. Light & G. Butterworth (Eds.), Context and cognition:
Ways of learning and knowing (pp. 28–46). Hillsdale, NJ: Erlbaum.
Messick, S. (1989). Validity. In R. L. Linn (Ed.), Educational measurement
(3rd ed., pp. 13–103). New York: Macmillan.
O’Malley, M. J., & Chamot, A. U. (1990). Learning strategies in second
language acquisition. Cambridge, England: Cambridge University Press.
Oxford, R. (1990). Language learning strategies: What every teacher should
know. Boston: Heinle & Heinle.
Oxford, R. (1993). La différence continue. . .: Gender differences in second/
foreign language learning styles and strategies. In J. Sutherland (Ed.),
Exploring gender (pp. 140–147). Englewood Cliffs, NJ: Prentice-Hall.
Oxford, R. (1996). Employing a questionnaire to assess the use of language
learning strategies. Applied Language Learning, 7, 25–45.
Oxford, R., Lavine, R. Z., Felkins, G., Hollaway, M. E., & Saleh, A. (1996).
Telling their stories: Language learners use diaries and recollective
studies. In R. Oxford (Ed.), Language learning strategies around the
world: Cross-cultural perspectives (pp. 19–34). Honolulu: University of
Hawai‘i, Second Language Teaching and Curriculum Center.
 Phakiti 689

Oxford, R., & Nyikos, M. (1989). Variables affecting choice of language

learning strategies by university students. Modern Language Journal,
73, 292–300.
Oxford, R., Nyikos, M., & Ehrman, M. (1988). Vı́ve la différence? Reflec-
tions on sex differences in use of language learning strategies. Foreign
Language Annals, 21, 321–329.
Oxford, R., Park-Oh, Y., Ito, I., & Sumrall, M. (1993). Learning a language
by satellite: What influences achievement. System, 21, 31–48.
Phakiti, A. (2003). A closer look at the relationship of cognitive and meta-
cognitive strategy use to EFL reading achievement test performance.
Language Testing, 20, 26–56.
Piaget, J. (1971). Biology and knowledge: An essay on the relations between
organic regulations and cognitive processes. Chicago: University of
Chicago Press.
Purpura, J. E. (1997). An analysis of the relationships between test takers’
cognitive and metacognitive strategy use and second language test per-
formance. Language Learning, 47, 289–325.
Purpura, J. E. (1999). Learner strategy use and performance on language tests: A
structural equation modeling approach. Cambridge, England: University of
Cambridge Local Examinations Syndicate and Cambridge University Press.
Read, J. (2000). Assessing vocabulary. Cambridge, England: Cambridge
University Press.
Salataci, R., & Akyel, A. (2002). Possible effects of strategy instruction on
L1 and L2 reading. Reading in a Foreign Language, 14, 1–17.
Scarcella, R., & Zimmerman, C. (1998). Academic words and gender: ESL
student performance on a test of academic lexicon. Studies in Second
Language Acquisition, 20, 27–49.
Schraw, G., & Moshman, D. (1995). Metacognitive theories. Educational
Psychology Review, 7, 351–371.
Sheorey, R. (1999). An examination of language learning strategy use in
the setting of an indigenized variety of English. System, 27, 173–190.
Shohamy, E. (2001). The power of tests: A critical perspective on the uses of
language tests. New York: Longman.
SPSS, Inc. (2001). SPSS Version 10.0 [Computer software]. Chicago: SPSS Inc.
Spurling, S., & Ilyin, D. (1985). The impact of learner variables on language
test performance. TESOL Quarterly, 19, 283–301.
Stevens, J. (1996). Applied multivariate statistics for the social sciences.
Mahwah, NJ: Erlbaum.
Vandergrift, L. (1997). The comprehension strategies of second language
(French) listeners: A descriptive study. Foreign Language Annals, 30,
387–406.
690 Language Learning Vol. 53, No. 4

Weinert, F. (1987). Metacognition and motivation as determinants of effective

learning and understanding. In F. Weinert & R. Kluwe (Eds.), Metacogni-
tion, motivation and understanding (pp. 1–16). Hillsdale, NJ: Erlbaum.
Weir, C. J. (1993). Understanding and developing language tests. London:
Prentice-Hall.
Wen, Q., & Johnson, R. K. (1997). L2 learner variables and English
achievements: A study of tertiary-level English majors in China. Applied
Linguistics, 18, 27–48.
Wesche, M., & Paribakht, T. S. (1996). Assessing second language vocabu-
lary knowledge: Depth vs. breadth. Canadian Modern Language Review,
53, 13–39.
Willingham, W. W., & Cole, N. S. (1997). Gender and fair assessment.
Mahwah, NJ: Erlbaum.
Wilson, J. (2000). Assessing metacognition. Unpublished doctoral disserta-
tion, University of Melbourne, Melbourne, Australia.
Yang, Y. (2002). Reassessing readers’ comprehension monitoring. Reading
in a Foreign Language, 14, 18–42.
Young, D. J., & Oxford, R. (1997). A gender-related analysis of strategies
used to process written input in the native language and a foreign
language. Applied Language Learning, 8, 43–73.
Zeidner, M. (1998). Test anxiety: The state of the art. New York and London:
Plenum Press.

Appendix A

Description of the Reading Comprehension Test

and Sample Questions

Objectives

Upon the completion of this course, students are able to (a) scan and
skim text for general and specific information; (b) find answers explicitly or
implicitly to questions; (c) recall word meanings; (d) skim to evaluate
information; (e) guess meanings of unknown words from context clues;
(f) identify phrases or word equivalence; (g) predict topics of passages and the
content of a passage from an introductory paragraph; (h) recognize abbre-
viations of words; (i) make decisions for appropriate information; (j) dis-
criminate between more or less important ideas; (k) distinguish facts from
opinions; (l) analyze reference words; (m) draw inferences from the content;
 Phakiti 691

(n) identify the title of the text and the appropriate heading; (o) summarize
the content of the given text; and (p) identify the main ideas or the purpose
of a passage.

Test Structure

Texts used in the test can be described as composed of (a) grammatical

features (e.g., the relationship between the structure of sentences and the
vocabulary used, such as sentence types and verb forms); (b) pragmatic
features (e.g., the principal intent of the writer, such as exposition and
argument); and (c) discourse features (e.g., the relationship between the
nature and the structure of the text as a whole), such as rhetorical properties
(e.g., definition, description, classification, illustration, cause/effect, problem/
solution and comparison/contrast) and textual organizations (e.g., documents
and prose). The topics on the test were related to the topics taught in the class,
such as family, food and drink, clothing, health, travels, and transportation.
The test was composed of two major sections.
Section 1: Gap-filling. The gap-filling or rationale cloze is a selec-
tive approach to the deletion of words from the text that allows both
lower-order (e.g., lexical choice of words and their interaction with other
words in context) and higher-order reading skills (e.g., pragmatic levels
and functions of sentences and the whole text structure) reading skills
(Alderson, 1979, 2000; Read, 2000). Section 1 of the test consisted of two
passages with 20 item gaps and two dialogue passages with 25 item
gaps for test takers to complete. It was designed to measure the test
takers’ ability to comprehend texts using both structural and lexical
appropriacy, pragmatics, and discourse. Items tested were selected
based on the structural and lexical appropriacy, pragmatics, and dis-
course taught in the course (e.g., appropriate use of verb forms and
functions, vocabulary).
There are three interrelated levels of meaning in text in the rationale
cloze (Read, 2000):

1. The local/micro level: This level focuses on the lexical choice of

words and their interaction with other words in a given text.

2. The pragmatic level: This level is extratextual and draws on lear-

ners’ general knowledge of the world.

3. The global/macro level: This level deals with the functions of the
sentences and the structure of the text as a whole.
692 Language Learning Vol. 53, No. 4

Section 2: Reading Comprehension. Section 2 consisted of four

passages, ranging from 139 words to 657 words. This section aimed to
measure the test takers’ ability to read English texts for main ideas,
details, inferences, etc., as defined in the objectives. A number of types of
questions were asked in this section:

Question Explanation Example

Main idea, These questions ask learners ‘‘What is the main idea
main topic, to identify an answer choice of this passage?’’
and main that correctly summarizes ‘‘What is the writer’s main
purpose the main idea and subject purpose in writing this
questions of the whole passage or the article?’’
author’s purpose in
writing the passage.
Factual These questions ask learners ‘‘According to the passage,
questions to locate and identify answers where is Mauritius
to questions about specific located?’’
information and details in ‘‘According to the passage,
the passage. dengue fever. . .’’
‘‘Which is true about the
dodo according to the
passage?’’
Negative These questions ask learners ‘‘Which in the passage is
questions which of the answer choices not true about. . .?’’
is not discussed in the ‘‘According to the passage,
passage. the following are benefits
of using electronics to
produce cars, except. . .’’
Inference These questions ask learners ‘‘The author implies which
questions to draw conclusions based of the following is true?’’
on information in the passage. ‘‘Which of the following
can be inferred from the
passage?’’
Vocabulary- These questions ask learners ‘‘The word ‘overlook’ (line 6)
in-context to identify the meaning of a can be best replaced by. . .’’
questions word or phrase as used in
the passage.
 Phakiti 693

Reference These questions ask learners ‘‘The word ‘it’ (line 7)

questions to identify the noun to refers to. . .’’
which a pronoun or
other expression refers.

Scoring Criteria

All the items on this test were equally important in representing the
reading comprehension achievement.

Test Sample

Final Examination
Time allowed: 3 Hours
Total Scores: 40%

General Directions
1. Write your name, student code, and major field of study on both the
examination paper and the answer sheet before you start doing the
test.

2. Read all the directions carefully and make sure that you understand
them. Ask the proctor(s) for clarification.

3. There are two parts to this section of the test: (1) Gap-Filling (45
Marks) and Reading Comprehension (40 Marks). Part 1 is 15% and
Part 2 is 25%.

4. There are 13 pages in this test including the cover sheet.

5. Answer all the questions choosing a, b, c, or d. There is only ONE

correct answer for each question. Therefore, do not select more than
ONE answer for each question because you will not get a mark on
that question. Answer them on your answer sheet. Put a cross (X)
on the correct answer. If you have made any changes, make sure
that you have chosen only ONE answer for each question.

6. Before you return the test and answer sheet to the proctor(s), please
answer the questionnaire regarding your thinking during the test.

7. If you are cheating in the examination, you will receive ‘‘F’’ auto-
matically in every subject you have been enrolled in this semester.
694 Language Learning Vol. 53, No. 4

Part 1: Gap-Filling
A. Directions: Read and complete the following passages choosing a, b, c,
or d.

Grilled Salmon Portions

Ingredients:
1. 6 salmon portions 7. 1 teaspoon freshly chopped ginger
2. 250 ml dry red wine 8. 1/2 teaspoon salt
3. 125 ml olive oil 9. 1/4 teaspoon tabasco
4. 125 ml lemon juice 10. 4 large capsicum rings
5. 2 shallots (chopped) 11. 4 large slices white onion
6. 2 teaspoons freshly crushed garlic

Method:
1. (1). . .. . .. . .Salmon under cold running water, pat dry, and
(2). . .. . .. . .aside.

2. In shallow glass bowl, combine all except the last two

(3). . .. . .. . .above. Place salmon in marinade. Cover and (4). . .. . .. . .for
1–2 hours.

3. Remove it from refrigerator and bring it to room temperature.

(5). . .. . .. . .it from marinade and reserve the liquid.

4. (6). . .. . .. . .Salmon, capsicum and onion on well-oiled barbecue (or

grill over medium heat) (7). . .. . .. . .6–7 minutes (or until lightly
browned).

5. Baste with marinade, turn, and (8). . .. . .. . .again. Continue cooking

for another 5 minutes (or (9). . .. . .. . .flesh is cooked properly).

6. To serve, (10). . .. . .. . .Salmon with capsicum and onions.

1. a. Boil b. Chop c. Cook d. Rinse

2. a. chop b. put c. serve d. set
3. a. dishes b. ingredients c. menus d. pieces
4. a. close b. open c. refrigerate d. top
5. a. Remove b. Roll c. Stir d. Wash
6. a. Boil b. Burn c. Cook d. Fry
7. a. for b. in c. on d. since
8. a. baste b. beat c. fold d. melt
 Phakiti 695

9. a. enough b. only c. still d. until

10. a. brown b. pour c. top d. wash

Part 2: Reading Comprehension

Directions: Read the following passages and choose the best answer a, b, c,
or d for each question.

Once Bitten

Everyone has been irritated by mosquitoes at one time or another. Thankfully, for most people
all the buzzing creatures leave behind are a few unpleasant bumps which itch for a few days. For
others, however, mosquito bites can lead to diseases such as malaria, dengue fever and
encephalitis, all of which can lead to death’s door.

5 Unfortunately, tropical Thailand is home to more than 400 types of mosquitoes. Fortunately,
only a few of these carry disease. ‘‘We can’t overlook these small creatures,’’ warned Dr. Seeviga
Saengtharatip. ‘‘They can be both a petty nuisance and life-threatening.’’ For city dwellers, the
ones to watch out for are the black-and-white striped, Aedes mosquitoes which carry Dengue fever.
At present, Dengue fever is one of the Kingdom’s top medical problems.

10 And the situation will become even more severe during the rainy season, as the rain water lying
around provides a perfect breeding ground for these flying pests.

The simplest way to avoid being bitten is to stay away from mosquitoes. Here is Dr. Seeviga’s
advice.
- Don’t go where mosquitoes live, especially during their peak feeding hours of early
15 morning and early evening. Avoid poorly-lit areas and dark corners.
- Because mosquitoes need still water to reproduce, they tend to live around swampy areas.
Thus, avoid walking or camping in these places.
- Wear protective clothes. Long-sleeved shirts and long pants will help keep mosquitoes from
biting you. Do not wear tight outfits as mosquitoes can penetrate even denim.

20 No matter how much time and money you spend killing and repelling mosquitoes, others will
always be back for more. Dr. Seeviga, therefore, suggested the best way to tackle the nuisance is
to eliminate the mosquitoes’ breeding grounds and kill them while they are still larva. Firstly,
empty standing water in old tyres, buckets, plastic covers, rain barrels, toys, or any other
container where mosquitoes larva can live. Secondly, keep fish in potted trays or lotus ponds,
25 and change the water in flower vases or jars at least once a week. Thirdly, clean guttering
regularly so rainwater won’t accumulate, and install screens on windows and doors. These
screens should be checked regularly for holes. If found, damage should be repaired immediately.
696 Language Learning Vol. 53, No. 4

Finally, use a sand-like material called ‘‘abate’’ in jars and vases containing water. The
chemical-coated substance releases a non-toxic chemical which kills the larvae, but leaves the
30 water safe for use, and even consumption.

(Adapted from ‘‘Once Bitten,’’ Bangkok Post, June 28, 1999)

(Sample questions only)

1. According to the passage, dengue fever. . .

a. is a type of mosquitoes.
b. is an incurable disease.
c. is carried by most mosquitoes.
d. outbreaks in the rainy season.

2. The word overlook can be best replaced by. . .

a. find
b. ignore
c. kill
d. stop

3. What does They refer to?

a. cities
b. doctors
c. mosquitoes
d. people

4. What is the main idea of Paragraph 4?

a. Always empty standing water in old tyres, buckets, toys, etc.

b. It is difficult to kill mosquitoes, especially in rainy seasons.
c. Mosquitoes live a long life in the rainy season.
d. The best method to prevent mosquitoes is to kill their larvae.
 Phakiti 697

Appendix B

Preliminary Cognitive and Metacognitive Strategy Taxonomy

(Based on O’Malley & Chamot, 1990, with Alderson, 2000; Baker &
Brown, 1984; Oxford, 1990; Purpura, 1999)

Cognitive and Metacognitive Strategies

Cognitive Strategies

1. Inferencing: Using available information within the reading texts to

guess the meaning of unfamiliar language items, to predict out-
comes, or to fill in missing information

1.1. Linguistic inferencing

Using known words in the reading passage to guess the meanings of

unknown words

1.2. Between-parts inferencing

Using available information beyond the local sentential level to

guess meaning

2. Elaboration: Using prior knowledge from outside the text and relat-
ing it to knowledge gained from the text in order to predict outcomes
and answer the questions correctly

2.1. Personal elaboration

Referring to prior personal experience

2.2. World elaboration

Using knowledge gained from experience in the real world

2.3. Academic elaboration

Using knowledge gained in academic situations such as from the

classroom
698 Language Learning Vol. 53, No. 4

2.4. Questioning elaboration

Using a combination of questions and world knowledge to brain-

storm logical possibilities of text meaning

3. Summarization: Making a mental or written summary of informa-

tion presented in reading tasks

4. Translation: Rendering ideas from one language to another in a

relative verbatim manner

5. Transfer: Using native language as a base to facilitate understand-

ing of reading in another language

6. Repetition: Repeating reading in the course of completing reading

tasks

7. Note-taking: Writing down main ideas, important aspects, outline,

key words, summary of the concepts to help complete reading tasks

8. Deduction/induction: Applying acquired or self-developed rules or

different words or phrases to accomplish reading tasks

9. Substitution: Selecting alternative approaches, revised plans, or

different words or phrases to accomplish reading tasks

Metacognitive Strategies

1. Planning: Previewing or overviewing the organization of the test

(including reading passages and test tasks); developing directions of
what needs to be completed; developing appropriate actions or strat-
egies to handle reading tasks; creating a plan for each reading
passage and its tasks to overcome difficulties that may interfere
with successful completion of the reading tasks

1.1. Advance preparation

Organizing a comprehensive preview to clarify the objectives, antici-

pate reading tasks, and pose strategies for handling them

1.2. Problem identification

Identifying what types of reading tasks or problems need to be

solved and selecting appropriate strategies to solve them
 Phakiti 699

1.3. Goal setting or selective attention

Deciding beforehand to attend to general and/or specific aspects of

language input or situational details that help one understand
reading tasks and task completion and ignore irrelevant information

1.4. Self-management

Understanding the conditions that help one successfully accomplish

reading tasks and arranging for the presence of those conditions

1.5. Goal prioritization

Determining which goals are more important and thus are neces-
sary to complete before others

2. Monitoring: Checking, verifying, or correcting reading performance

against external standards while or after completing reading tasks

2.1. Comprehension monitoring

Checking, verifying, or correcting understanding of or mistakes in

comprehending the text and completing questions at local and global
levels

2.2. Double-checking

Checking, verifying, or correcting understanding across the tasks a

second (or more) time

2.3. Performance evaluation

Evaluating overall performance of the reading tasks, that is,

whether or not the required standards or set goals have been
reached

2.4. Strategy monitoring and evaluation

Evaluating the effectiveness of current strategies being used

2.5. Problem monitoring and evaluation

Explicitly identifying the central point that needs resolution in a

task or identifying an aspect of the task that hinders its successful
completion
700 Language Learning Vol. 53, No. 4

Appendix C

Cognitive and Metacognitive Questionnaire

Directions: A number of statements which people use to describe

themselves when they were taking a reading test are given below. Read
each statement and indicate how you thought during the test. Choose 1
(Never), 2 (Sometimes), 3 (Often), 4 (Usually), and 5 (Always).

Your thinking 1 2 3 4 5
1. I made short notes or underlined main 1 2 3 4 5
ideas during the test.
2. I translated the reading texts and tasks 1 2 3 4 5
into Thai.
3. I used pictures or titles of the texts 1 2 3 4 5
to help comprehend reading tasks.
4. I used my own English structure 1 2 3 4 5
knowledge to comprehend the text.
5. I spent more time on difficult questions. 1 2 3 4 5
6. I tried to understand the texts and 1 2 3 4 5
questions regardless of my vocabulary
knowledge.
7. I tried to find topics and main ideas by 1 2 3 4 5
scanning and skimming.
8. I read the texts and questions several 1 2 3 4 5
times to better understand them.
9. I used my prior knowledge to help 1 2 3 4 5
understand the reading test.
10. I tried to identify easy and difficult 1 2 3 4 5
test tasks.
11. I looked at the scores of each part to 1 2 3 4 5
determine the weight of scores before
starting to complete the test.
12. I determined which parts were more 1 2 3 4 5
important than others before starting
the test.
13. When I started to complete the test, 1 2 3 4 5
I planned how to complete it and
followed the plan.
14. I was aware of what and how I was 1 2 3 4 5
doing in the test.
 Phakiti 701

15. I checked my own performance and 1 2 3 4 5

progress while completing the test.
16. I attempted to identify main points 1 2 3 4 5
of the given reading texts and tasks.
17. I thought through the meaning of 1 2 3 4 5
the test tasks/questions before
answering them.
18. I was aware of which strategy to 1 2 3 4 5
use and how and when to use it.
19. I corrected mistakes immediately 1 2 3 4 5
when found.
20. I asked myself how the test questions 1 2 3 4 5
and the given texts related to what I
already knew.
21. I determined what the test tasks/ 1 2 3 4 5
questions required me to do.
22. I was aware of the need to plan 1 2 3 4 5
a course of action.
23. I was aware of how much of the test 1 2 3 4 5
remained to be completed.
24. I tried to understand the questions 1 2 3 4 5
adequately before attempting to find
the answers.
25. I made sure I understood what had 1 2 3 4 5
to be done and how to do it.
26. I was aware of my ongoing thinking 1 2 3 4 5
processes.
27. I kept track of my own progress to 1 2 3 4 5
complete the questions on time.
28. I used multiple thinking strategies 1 2 3 4 5
to help answer the test questions.
29. I made sure to clarify the goal and 1 2 3 4 5
know how to complete it.
30. I was aware of the selected strategies 1 2 3 4 5
to help me complete the test questions
before solving them.
31. I checked my accuracy as I progressed 1 2 3 4 5
through the test.
32. I selected relevant information to 1 2 3 4 5
help me understand the reading texts
and answer the test questions.
702 Language Learning Vol. 53, No. 4

33. I determined how to solve the test. 1 2 3 4 5

34. I carefully checked the answers before 1 2 3 4 5
submitting the test.
35. I thought about how I had completed 1 2 3 4 5
the test.