Вы находитесь на странице: 1из 23

S3wtem. Vol. 23, No. I, pp.

1-23, 1995
Copyright 1995 Elsevier Science Ltd
~ Pergamon Printed in Great Britain. All rights reser,,ed
0346-25 IX/95 $9.50+0.00

0346-251X(94)00047-6

ASSESSING THE USE OF LANGUAGE LEARNING STRATEGIES WORLDWIDE


WITH THE ESL/EFL VERSION OF THE STRATEGY INVENTORY FOR
LANGUAGE LEARNING (SILL)

REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

University of Alabama, Tuscaloosa, AL, USA

With factor analysis contributions by Neff Anderson, Ohio University, USA;


Deena Boraie, American University in Cairo, Egypt; John Green, University of
Puerto Rico at Mayaguez and Salem State College, USA; Gene Halleck, Oklahoma
State University, USA; Omneya Kassabgy, Career Development Center, Cairo,
Egypt; Victoria Talbott, Skagit Valley Junior College, USA; Yoshinori Watanabe,
Japan; Nae-Dong Yang, National Taiwan University, Taiwan, ROC; Wenpeng
Zhang, Ohio University, USA

Summative rating scales are among the most efficient and comprehensive ways to
assess frequency of language learning strategy use. This article discusses
applications of this assessment technique and describes the most widely employed
strategy scale, the ESL/EFL version of the Strategy Inventory for Language
Learning (SILL). Reliability of the SILL is high across many cultural groups.
Validity of the SILL rests on its predictive and correlative link with language
performance (course grades, standardized test scores, ratings of proficiency), as
well as its confirmed relationship to sensory preferences. Studies of strategy use
frequencies and factor analytic results across cultures are included.

INTRODUCTION

One of the most prevalent ways to assess the use of language learning strategies is to use a
summative rating scale, popularly known as a questionnaire, an inventory, or (less accurately)
a survey. The most often used strategy scale around the world at this time is the Strategy
Inventoryfor Language Learning (SILL, Oxford, 1986-1990). This article has four purposes:
(1) to present information on the advantages and disadvantages of using a strategy scale in
comparison with other means of strategy assessment; (2) to discuss strategy scales other than
the SILL; (3) to provide detailed results concerning the ESL/EFL SILL itself: utility,
reliability, validity, frequency-of-strategy-use studies, and underlying factor structure; and (4)
to offer implications for research and instruction.
2 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

ADVANTAGES AND DISADVANTAGES OF STRATEGY INSTRUMENTS

Compared with the other strategy assessment techniques, student-completed, summative rating
scales have a number of advantages. These self-report scales are easy and quick to give,
provide a general assessment of each student's typical strategies across a variety of possible
tasks, may be the most cost-effective mode of strategy assessment, and are almost completely
nonthreatening when administered using paper and pencil (or computer) under conditions of
confidentiality. Moreover, many students discover a great deal about themselves from taking a
strategy scale, especially one like the SILL that is self-scoring and that provides immediate
learner feedback. However, a disadvantage of the SILL and other strategy scales is that they
do not describe in detail the language learning strategies a student uses in response to any
specific language task (as does the more time-consuming think-aloud protocol).

Each of the other techniques also has advantages and disadvantages (explored in more detail
by Cohen, 1987; Oxford, 1990b). For example, informal and formal observation are easy to
use in the classroom but cannot provide information on unobservable, mental strategies such
as reasoning or analyzing. Interviews, whether formal or not, provide personalized information
on many types of strategies that would not be available through observation, but they take
considerable time from the teacher and the students. Group discussions can give a wonderful
picture of the strategies used by the class as a whole, but they do not offer full information
about the strategies used by any individual student. Language learning diaries and dialogue
journals provide detailed, rich data on learning strategies for individuals, but the data do not
provide direct comparisons between students because of the open-ended nature of the diaries
or journals. Recollective narratives (or other recollective modes) generally unite language
learning strategies with other important aspects of learning, such as motivation and learning
style, providing a "big picture" of the whole learning process, yet recollectives might (in
learners whose memories tend toward leveling rather than toward sharpening) be influenced
by slight loss of detail. Think-aloud protocols offer the most detailed information of all
because the student describes strategies while doing a language task; but these protocols are
usually used only on a one-to-one basis, take a great deal of time, reflect strategies related just
to the task at hand (not a general portrait of the individual's strategies in toto), and are not
summative across students for group information.

Most of these strategy assessment techniques involve some type of learner self-report. The
reason for researchers' frequent use of learner self-report is that it is often difficult for
researchers to employ standard observational method. Thus, much of the research on language
learning strategies depends on learners' willingness and ability to describe their internal
behaviors, both cognitive and affective (emotional), as noted by Oxford (1990b) and Harlow
(1988). This situation has led some people to question learning strategy research because of
possible problems in self-reporting: "social desirability" biases in responses, over-subjectivity,
inability to verbalize clearly, and low self-awareness among certain learners. Nevertheless,
researchers have discovered, through conducting repeated studies with clear instructions in
situations where no grades or sanctions are involved with strategy use, that many or most
language learners are capable of remembering their strategies and describing them lucidly and
in a relatively objective manner (see, e.g. Chamot and Kupper, 1989; O'Malley and Chamot,
1990).
STRATEGY INVENTORY FOR LANGUAGE LEARNING 3

STRATEGY RATING SCALES OTHER THAN THE SILL

Bialystok (1981) used a 12-item, structured, untitled rating scale to assess strategy use. The
scale asked questions about the extent to which strategies were used on both oral and written
tasks in communicative settings (the strategies were functional practice and inferencing or
guessing) and in formal classroom settings (the strategies were formal practice and
monitoring). Using the scale with students of French in grades 10 and 12 in Canada, Bialystok
found that functional practice had a stronger relationship with achievement than did any of the
other strategies, even though monitoring and inferencing were used more often. Formal
practice with rules and structures was less effective as students advanced to higher levels of
learning, but functional practice had no such limitation. Reliability and validity data were
absent for this instrument.

Politzer (1983) published 1-4-scaled strategy scale including 51 items divided into three
groups: general behaviors, classroom behaviors, and interactions outside of class. Using this
survey with US university students of French, German, and Spanish, Politzer found that
course level influenced strategy use, with higher-level students using more so-called
"positive" strategies (i.e. strategies related to communicative language proficiency); and that
females used social learning strategies more often than males. No reliability or validity data
were given.

Politzer and McGroarty (1985) used a somewhat similar Behavior Questionnaire containing
66 items divided into three groups: individual study behaviors, classroom behaviors, and
interactions outside of class. Reliability was marginally acceptable (.51, .61, and .63). The
survey was used with students learning intensive ESL in an eight-week course. Improvements
in ESL achievement were related to individual strategies, such as asking questions for
clarification. Successful strategies for grammar differed from those for listening and speaking.
Major academic field had a significant effect on strategy choice, with engineers avoiding
strategies that were deemed "positive" for gaining communicative language proficiency; but
there was an overlap with nationality, since many engineers were also Asian.

McGroarty (1987) used a 56-item Language Learning Strategy Student Questionnaire with a
0-6 range, divided into the same three groups as in the Politzer and McGroarty study above.
No reliability or validity data were published. University students of Spanish, although taught
by communicative methods, nevertheless avoided authentic practice strategy and used
traditional learning strategies, such as relying heavily on the dictionary.

The Learning Strategies Inventory (Chamot et al., 1987) is a 48-item, 1-4-scaled instrument
divided into five parts: listening in class, speaking in class, listening and speaking outside of
class, writing, and reading. The items showed different ways of applying a total of 16
strategies. Students of Russian used more strategies than students of Spanish. Spanish and
Russian students used somewhat different strategies across language levels (beginning and
intermediate/advanced). No data were published on reliability or validity.

Padron and Waxman (1988) developed a 14-item, 1-3-scaled instrument to assess reading
strategies of Hispanic ESL students in grades 3-5. Seven of the items were expected to be
positively related to learning and seven negatively related. Results showed that six of the
4 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

seven most-used strategies were in the predicted-positive group. However, only two strategies
were significantly related to learning outcomes, and these were both in the negative direction;
no strategies significantly helped learning to occur. No reliability or validity data were offered.

Bedell (1993) points out a number of additional strategy scales. Huang (1984) and Huang and
van Naerssen (1987) used a Strategies Questionnaire for Chinese EFL learners. This
instrument includes some scaled items and some yes-no items, as well as free-response
questions. Most of the items concern strategies for improving listening and speaking skills.
Wangsotorn et al. (1986) used the Chulalongkorn University Language Institute Learning
Strategy Form A (consisting of 42 yes-no statements about students behaviors) for Thai
learners of EFL. Kim (1991) designed a Perceptual Learning Strategy Questionnaire,
including 18 items. Noguchi's (1991) Questionnaire for Learners is an instrument with 24
items on a 3-point scale and 24 more on a 4-point scale, based largely on items from the SILL.
Wen and Johnson's (1991) strategy scale is also adapted from the SILL.

Few of the above instruments have any published reliability or validity data. This is the key
reason that the SILL was developed. If the psychometric properties of reliability and validity
have not been explored, it is impossible to know whether we can put faith in the results of the
research. Another reason for developing the SILL is that the instruments just mentioned do not
always systematically represent all the kinds of strategies viewed as important to language
learning. A more comprehensive scale was needed for measuring strategy use among ESL and
EFL students.

THE STRATEGY INVENTORY FOR LANGUAGE LEARNING (SILL)

Development
The SILL (Oxford 1986-present) was first designed as an instrument for assessing the
frequency of use of language learning strategies by students at the Defense Language Institute
in Monterey, California. Two revised versions of the SILL, one for foreign language learners
whose native language is English (80 items) and the other for learners of English as a second
or foreign language (ESL/EFL, 50 items), were published in an appendix to Oxford's (1990b)
learning strategy book for language teachers. This article deals only with research done using
the 50-item (short) version.

It is estimated that 40-50 major studies, including a dozen dissertations and theses, have been
done using the SILL. These studies have involved an estimated 8000-8500 language learners.
According to research reports and articles published in the English language within the last
10-15 years, the SILL appears to be the only language learning strategy instrument that has
been extensively checked for reliability and validated in multiple ways.

The SILL uses a choice of five Likert-scale responses for each strategy described: never or
almost never true of me, generally not true of me, somewhat true of me, generally true of me,
and always or almost always true of me. The SILL response options were based on the widely
used and well accepted response options of the Learning and Study Strategies Inventory
described by Weinstein et al. (1987). On the SILL, learners are asked to indicate their
response (1, 2, 3, 4, or 5) to a strategy description, such as "I try to find patterns in English" or
STRATEGY INVENTORY FOR LANGUAGE LEARNING 5

"I plan my schedule so I will have enough time to study English." In addition to the original
English version, the ESL/EFL SILL has been translated into the following languages: Arabic,
Chinese, French, German, Japanese, Korean, Russian, Spanish, Thai, and Ukrainian.

In 1989, the SILL was organized according to strategy groups using a statistical procedure
called factor analysis. This procedure allows the researcher to subdivide an instrument into
dimensions usually referred to subscales or factors. Six subscales were developed based on the
early factor analyses, with the intent that each subscale would have an adequate number of
items to facilitate more in-depth research and understanding of the learning strategies for
ESL/EFL. These early subscales include:

(1) Memory.strategies, such as grouping, imagery, rhyming, and structured reviewing (nine
items).
(2) Cognitive strategies, such as reasoning, analyzing, summarizing (all reflective of deep
processing), as well as general practicing (14 items).
(3) Compensation strategies (to compensate for limited knowledge), such as guessing
meanings from the context in reading and listening and using synonyms and gestures to
convey meaning when the precise expression is not known (six items).
(4) Metacognitive strategies, such as paying attention, consciously searching for practice
opportunities, planning for language tasks, self-evaluating one's progress, and monitoring
error (nine items).
(5) Affective (emotional, motivation-related) strategies, such as anxiety reduction, self-
encouragement, and self-reward (six items).
(6) Social strategies, such as asking questions, cooperating with native speakers of the
language, and becoming culturally aware (six items).

As shown above, the largest group of items is the cognitive strategies. This stands to reason,
because research on learning strategies suggests that cognitive strategies possess the greatest
variety, covering strategies related to practice and to the all-important "deep processing" in
which learners analyze, synthesize, and transform new information (Oxford and Ehrman,
1995).

A SILL package includes: a short set of directions to the student with a sample item, the
50-item instrument, a scoring worksheet on which students record their answers and calculate
their averages for each strategy subscale and their overall average, a summary profile that
shows students' results and provides examples for student self-interpretation, and a strategy
graph that allows each learner to graph results from the SILL. A background questionnaire is
also available to document age, sex, language experience, motivation, and other information
(see Oxford, 1990b).

PSYCHOMETRIC QUALITIES OF THE ESL/EFL SILL

This section describes the psychometric quality of the 50-item ESL/EFL SILL. Normally,
such quality is established and presented in terms of utility, reliability, and validity. (Note that
psychometric quality data are also available for the longer form of the SILL that was designed
6 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

for native English speakers learning foreign languages; see Oxford, 1992; Oxford and
Ehrman, 1995.)

Utility
An instrument might be reliable and valid (see explanations below) without being very useful.
The utility of an instrument is important to its overall psychometric quality. Utility is the
usefulness of an instrument in real-world settings for making decisions relevant to people's
lives. The SILL has utility, according to the many people around the world who have
employed it. The most frequent venue has been the classroom, where the goal has been chiefly
to reveal the relationship between strategy use and language performance. The reason this goal
is important is that if there is a strong relationship between these two variables, perhaps
language performance can be improved by enhancing strategy use. Other classroom uses of
the SILL have included assessing strategy use at a given point, to be compared with strategy
use later (sometimes after strategy training interventions); comparing the often very different
learning strategies of women and men; making the conceptual linkage between strategy use
and underlying learning styles; and individualizing classroom instruction based on the strategy
use of different students. So far the utility of the SILL has not included making placements of
individuals into language classes on the basis of strategy use results; that is a use the author of
the SILL does not particularly recommend. See the reference list for dozens of studies
showing various uses of the SILL.

Reliability
Reliability refers to the degree of precision or accuracy of scores on an instrument. In the case
of the SILL, Cronbach alpha, a measure of internal consistency, was chosen as the most
appropriate reliability index. Cronbach alpha reliability coefficient is used on continuous data
such as the Likert-type scale in the SILL.

Though the current ESL/EFL SILL was constructed using six subscales, reliability of the SILL
is determined with the whole instrument. In general, the ESL/EFL SILL reliabilities have been
high. With the ESL/EFL SILL, Cronbach alphas have been: .94 using the Chinese translation
with a sample of 590 Taiwanese university EFL learners (Yang, 1992a); .92 using the
Japanese translation with 255 Japanese university and college EFL students (Watanabe, 1990);
.91 using the Korean translation with 59 Korean university EFL learners (Oh, 1992); .93 using
the researcher-revised Korean translation with 332 Korean university EFL learners (Park,
1994); and .91 using the Puerto Rican Spanish translation with 374 EFL learners on the island
of Puerto Rico. (These reliabilities are similar to those--.91-.95--found for the 80-item
foreign language SILL given in the native language of the respondent; see Oxford, 1986;
Oxford and Nyikos, 1989; Wildner-Bassett, 1992a; Bedell, 1993; Nyikos and Oxford, 1993;
Oxford and Burry, 1993.)

Slightly lower but still very acceptable reliabilities are found for the ESL/EFL SILL when it is
not administered in the native language of the respondents but is given in English instead. All
the reliabilities below refer to heterogeneous (multi-language) groups of ESL learners in the
U.S. Phillips' (1990, 1991) data had a reliability of .87 with 141 students. SILL data from
Oxford et al. (1989) showed a reliability of .86 with 159 students. Anderson's (1993) data on
95 students had a reliability of .91. Involving 31 learners, Talbott's (1993) data had a
STRATEGY INVENTORY FOR LANGUAGE LEARNING 7

reliability of .85. A three-study combination (merging ESL data from Oxford et al., 1990;
Anderson 1993; Talbott 1993) showed a reliability of .88 with 137 students. Thus, reliability
of the ESL/EFL SILL goes down, but not greatly, when the SILL is administered in the target
language, English, rather than in the respondent's native language. The reliability of the SILL
administered in this manner contains somewhat more measurement error due to the
confounding language effect. However, these reliabilities are very respectable, and the SILL
can be administered in the respondent's native language or a foreign or second language with
confidence that measurement error is minimal.

Validio,
Validity refers to the degree to which an instrument measures what it purports to measure.
Several bases exist for validity: content validity, criterion-related validity, and construct
validity.

A content-validity study. Content validity is determined by professional judgement. The


content validity of the SILL is very high. Two strategy experts matched the SILL items with
agreement at .99 against entries in a comprehensive language learning strategy taxonomy,
which itself was built from a detailed blueprint of a range of over 200 possible strategy types
(for complete details see Oxford, 1986).

Criterion-related validity studies related to language performance. Criterion-related validity


involves either predictive or concurrent relationships between the key variable, in this case
learning strategy use, and other important variables, in this case language performance.
Predictive validity, one form of criterion-related validity, is established with the use of a
criterion and at least one predictor variable in a simple or multiple regression analysis.
Concurrent validity, another form of criterion-related validity, is demonstrated when data are
collected for all variables at one time.

Both concurrent and predictive SILL validity are shown in relationships between the SILL on
the one hand and language performance on the other. This evidence is probably the strongest
support possible to the assertion of the validity of the SILL. A number of studies have
demonstrated this relationship. In these ESL/EFL SILL studies, language performance is
measured in various ways: general language proficiency tests (Rossi-Le, 1989; Phillips, 1990,
1991; Chang, 1991; Wen and Johnson, 1991; Green and Oxford, 1992; Park 1994), oral
language proficiency tests (Chang, 1991), grades in a language course (Mullins, 1991),
language achievement tests directly related to course content (Oxford et al., 1993a, b),
proficiency self-ratings (Oxford and Nyikos, 1989; Watanabe, 1990; Chang, 1991), and
professional language career status (Ehrman and Oxford, 1989).

Here are some examples of the relationship between strategy use on the ESL/EFL SILL and
language performance. Rossi-Le (1989) found that for 147 adult ESL students in the
midwestern and the northeastern parts of the US, language proficiency level (on a standardized
test) predicted strategy use in multiple regression analyses. More proficient ESL students used
self-management strategies like planning and evaluating (p < .006) and formal practice
(p < .02) significantly more often than less proficient ESL students.
8 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

Strategy use was related to language achievement scores (final test grades) in a study
involving 107 high school students of Japanese. The ESL/EFL SILL was modified slightly for
the distance education students in this study by Oxford et al. (1993a, b). In a multiple
regression analysis, learning strategy use was a moderate but significant predictor of Japanese
language achievement (.20, p < .04). The only other significant predictor was the degree of
learner motivation (.30, p < .003).

Using a modified version of the ESL/EFL SILL translated into Japanese, Wen and Johnson
(1991) studied the learning strategies of 242 second-year English majors at seven
postsecondary institutions in Nanjing and Shanghai. These subjects had recent national
English proficiency scores that averaged 10 points higher than the country's mean. Using a
partial least square procedure, a multiple regression procedure, the researchers found that one-
third of the variance in English proficiency was related to combined effects of six variables,
three of which were groups of strategies taken from the SILL.

Takeuchi (1993a) used multiple regression and found that eight SILL items predicted 58% of
the variance in scores on the Comprehensive English Language Test. The CELT was used in
that study to measure English achievement among 78 Japanese first-year students of English at
a women's college in Kyoto. The figure of 58% is unusually high when compared with
previous studies. Four strategies positively predicted language achievement: writing notes,
messages, letters, or reports in English; trying not to translate word-for-word; dividing words
into parts to find meaning; and paying attention when someone is speaking English. Four
strategies negatively predicted language achievement: asking questions in English; using
flashcards; writing down feelings in a language learning diary; and trying to find as many
ways as possible to use English. Takeuchi explained some of these findings based on cultural
influences (see also Takeuchi, 1991a, b, 1993b).

Construct Validity Studies. Construct validity concerns how well a theoretical construct is
measured. A variety of statistical procedures is used to bring meaning to construct validity.
Since most instruments, including the SILL, are written to define an abstract notion of a
theoretical construct, it is necessary to use a number of statistical procedures to establish
evidence that this theoretical construct is defined by the items on the instrument. For instance,
correlation coefficients are an index from -1 to +1 which describes the degree to which two
phenomena "vary together" in strength (stronger is close to 1) and direction (positively or
negatively). Analysis of variance (ANOVA) and multivariate analysis of variance
(MANOVA) use pre-established groups to demonstrate construct validity. If two distinct
groups have different results on an assessment, the instrument differentiates between known
groups, which provides evidence of construct validity. Factor analysis and multidimensional
scaling are also used to document the dimensionality of the theoretical construct.

Watanabe (1990) asked university and college EFL students in Japan to rate from low to high
their own proficiency in English. These proficiency self-ratings correlated moderately (average
r = .30) with SILL strategies (p < .0005-.001), except for those in the social/affective
strategies. This means that in general most SILL strategies were used more often by students
who rated their language proficiency higher and used less often by students who rated their
language proficiency lower.
STRATEGY INVENTORY FOR LANGUAGE LEARNING 9

Chang (1991) used the SILL to investigate the learning strategies and English proficiency of
50 mainland Chinese and Taiwanese ESL students at a southeastern university in the US.
Three measures of proficiency (self-ratings and two standardized tests) showed different
effects on strategy use. Students who rated themselves above average in proficiency used more
strategies overall than those who rated themselves below average. Neither the scores on the
Test of English as a Foreign Language (TOEFL) nor the llyin Oral Interview significantly
affected overall strategy use, but students with high scores on the oral interview used
significantly more social strategies than those with low scores.

Park (1994) employed the SILL to determine the relationship between strategy use and
proficiency among 332 students of EFL at the Korea Maritime University and Inha University.
Park divided the subjects into three groups according to their strategy use: low, medium and
high. Then Park calculated TOEFL scores for each group. According to ANOVA, the TOEFL
mean scores of these three groups differed significantly from each other. Post hoc tests
showed that the high strategy group has a language proficiency score that was significantly
higher than that of the medium strategy use group, which in turn had a higher language
proficiency score than that of the low strategy use group. Thus, a linear relationship was
shown between strategy use and language proficiency. In addition, Park found that the
correlation between total TOEFL scores and strategy use was r = .34 (p < .0001). Cognitive,
social and metacognitive strategies had a higher relationship (r = .33, .30, and .28,
respectively) to TOEFL scores than did other kinds of strategies (memory, r = .24; affective,
r = .23; compensation, r = ,21).

Phillips (1990, 199 l) found strong relationships between ESL/EFL SILL frequencies and
English proficiency levels (measured by the TOEFL) among 141 adult ESL learners in seven
western states in the US. She found no consistent differences between high-proficiency
students and low-proficiency students on entire strategy categories, so she looked at strategies
singly. She found that middle scorers on the TOEFL, who thus had moderate proficiency in
English, showed significantly higher overall strategy use than did the high-proficiency or the
low-proficiency group, when strategy use was defined as the mean number of strategies used
frequently and the mean number of strategy categories that had at least one frequently used
strategy. The profile of medium-proficiency students using more strategies more often than
high-proficiency or low-proficiency students produced a curvilinear pattern. Additionally,
Phillips discovered that high TOEFL scorers used certain learning strategies significantly more
often than low TOEFL scorers: paraphrasing, defining clear goals for learning English, and
avoiding verbatim translation. The low TOEFL scorers reported significantly greater use of
certain strategies, many of which would logically be found among beginning students: using
flashcards, finding out how to be a better speaker, looking for conversation partners, noticing
tension or nervousness, and writing down feelings in a journal.

Green (1991) investigated 213 Spanish-speaking students learning English on the island of
Puerto Rico. These students can be designated as neither ESL students nor EFL students but
are instead a hybrid of the two categories owing to their high English input and their low
English output; they have great amounts of stimulation in English but do not need to produce
the language for survival reasons. The English as a Second Language Achievement Test
(ESLAT), which was used in the study, is a measure of overall English proficiency (not
achievement on a given curriculum). Green found moderate and significant correlations,
10 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

usually in the high .30s, between SILL strategy factors and ESLAT proficiency scores, and he
discovered the same level of correlations between individual SILL items and proficiency
scores. In a later analysis of variance, Green (1992):showed that language level had a
statistically significant influence on strategy use, with higher-proficiency students in general
using strategies more frequently than lower-proficiency students. With a larger sample of 374
students, Green and Oxford (in preparation) found that language proficiency level had
significant effects on the use of the following kinds of strategies: compensation strategies
(p < .0001), cognitive strategies (p < .0001), metacognitive strategies (p < .0025), and social
strategies (p < .008). Two other categories of strategies, memory and affective strategies,
displayed no significant difference by proficiency level. In the four significant categories,
higher proficiency was associated with more frequent strategy use. Significant variation
occurred by gender, with females using strategies significantly more often than males in this
study.

In Mullins' (1991) SILL study, 110 Thai university-level EFL majors showed linkages
between strategy use and various measures of English proficiency. For instance, compensation
strategy use correlated at r = .38 (p < .0001) with language placement scores and at r = .32 (p
< .006) with language course grades. A correlation of r = .24 (p < .03) was found between
metacognitive strategy use and language course grades. However, a negative correlation of
r = -.32 (p < .005) was found between affective strategy use and language entrance
examination scores, which are different from language placement scores in this particular Thai
university. It is possible that students who are very anxious and who resort to affective
strategies do less well on the entrance examination.

As shown by Dreyer and Oxford (in preparation), approximately 45% of the total variance in
language proficiency (TOEFL scores) in a South African ESL study was explained by learning
strategy use as measured by SILL. A regression analysis demonstrated that the greatest part of
the variance was accounted for by metacognitive strategies, with much smaller amounts
contributed by affective and social strategies. Canonical correlation showed a highly
significant relationship between the parts of the TOEFL and the categories on the SILL
(r = .73). The sample consisted of 305 Afrikaans first-year university students learning ESL in
South Africa (Dreyer, 1992).

Thus, ESL/EFL SILL strategy frequency is related, as expected, to language performance in a


number of studies, thus providing validity evidence for the SILL as a strategy instrument.
(These results agree with earlier research using varied strategy assessment instruments; for
instance, Huang, 1984; Corrales and Call, 1989; Oxford and Nyikos, 1989; O'Malley and
Chamot 1990.) In most but not all instances, the relationship is linear, showing that more
advanced or more proficient students use strategies more frequently.

Criterion-related and construct validity in relationship to learning styles. Strong relationships


between learning strategy use and sensory preferences---often viewed as an aspect of learning
style--have been posited (see, e.g. Oxford et al., 1991). Visual students are described by
Oxford et al. as using strategies involving reading alone in a quiet place or paying attention to
blackboards, movies, computer screens, and other forms of visual stimulation. Auditory
students are comfortable without visual input and frequently use strategies that encourage
conversation in a noisy, social environment with multiple sources of aural stimulation.
STRATEGY INVENTORY FOR LANGUAGE LEARNING 11

Kinesthetic students need movement strategies, and tactile students require strategies that
involve manipulating real objects in the classroom; both types need to use the strategy of
taking frequent breaks.

Some ESL/EFL SILL data exist supporting the link between learning strategy use and sensory
preferences, thus at the same time strengthening the evidence of validity of the SILL. Rossi-Le
(1989) found a significant relationship (p < .0005) between sensory preference (visual,
auditory, tactile and kinesthetic) and overall strategy use on the ESL/EFL SILL through a
MANOVA, and she also found significant predictive relationships through multiple
regression.

Rossi-Le's MANOVA results showed that visual learners preferred visualization strategies
(p < .0005). Auditory-style learners used memory strategies more than did other learners
(p < .0005). Compared with others, tactile learners demonstrated significant use of strategies
for searching for and communicating meaning (p < .006) and self-management/metacognitive
strategies (p < .02). Kinesthetic learners did not use general study strategies (p < .003) or self-
management/metacognitive strategies (p < .02) as often as others did.

The regression results indicated that a visual learning style predicted using visualization
strategies (B = .33, p < .00005). Being a visual learner, however, negatively predicted using
independent strategies (13 = -.22, p < .001), affective strategies (B = -.23, p < .009), and
strategies for searching for and communicating meaning (B = -.22, p < .008). Having an
auditory learning style significantly predicted memory strategies (13 -- .38, p < .0008) and self-
management or metacognitive strategies (13 = -.20, p < .01) but was a negative predictor of
employing authentic language use strategies (13 = -.20, p < .01). Being a tactile learner
significantly predicted employing authentic language use strategies (B = .26, p < .001) and
strategies for meaning (B = .32, p < .0002) but negatively predicted use of memory strategies
(B = -. 16, p < .04). A kinesthetic learning style predicted infrequent use of general study
strategies (13 = -.32, p < .002). Thus, these predictions are low-to-moderate and significant.

Fakability (social desirability) results: another indicator of validity. If people are not honest
in their answers, validity is destroyed. Social desirability response bias is the tendency of a
person to answer dishonestly for one of two purposes: to please the researcher or to show
himself or herself as being a good or socially acceptable person. This response bias can lead to
"faking the results". Is the ESL/EFL SILL likely to draw upon this kind of bias? Is the SILL
vulnerable to validity problems caused by dishonest responses? Yang (1992b, 1993) tested the
E S L / E F L SILL for "fakability" of responses (using the well known and respected
Marlowe-Crown Social Desirability Scale) with 505 Taiwanese students of English as a
foreign language. She found no statistical evidence that students faked any of their answers on
the SILL. In repeated studies with the ESL/EFL SILL, the first author of this article has
examined the data statistically to determine the number of low, medium, and high frequencies.
Social desirability response bias would cause respondents to try to show very frequent (very
high) use of strategies. This was not the case; the responses showed a range of strategy use and
were in no way clustered at the extremely high end (except for the responses of language
teachers, who might be expected to use significant numbers of strategies frequently).
12 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

STRATEGY FREQUENCY STUDIES USING THE EFL/ESL SILL

Frequency of use of language learning strategies appears to be directly related to whether


students are in an ESL or EFL setting (or in a hybrid of ESL and EFL environments, as in
Puerto Rico). Moreover, significant differences related to career interests, institution, cultural
background, and gender have been found in the frequency of strategy use. A few comparisons
of SILL frequencies are presented here. Averages of 3.5-5.0 were usually considered high
strategy use; 2.5-3.4 were designated medium strategy use; and 1.0-2.4 were regarded as low
strategy use.

Frequency o f strategy use related to foreign vs. second language environments


Oxford et al. (1989) found high frequencies of use for 60% of the strategies on the SILL with
159 ESL learners in the US. These strategies included self-management, authentic language
use, strategies for searching for and communicating meaning, general study strategies,
affective strategies, and formal practice strategies. Medium-frequencies of use in that study
were found for memory strategies, social strategies, visualization strategies, and strategies for
studying or practicing independently. Likewise, Rossi-Le (1989) found that among 147 adult
learners in two community colleges in the US, one in the midwest and the other in New
England, high frequencies of strategy use existed for most of the strategies, including social
strategies, authentic language strategies, visualization strategies, formal practice strategies,
metacognitive strategies, memory strategies, and affective strategies. The rest of the strategies,
including general study strategies, strategies for searching for and communicating meaning,
and strategies for studying independently, showed medium use. Similarly, Oxford et al.
(1989), with a sample of 43 ESL students at a large northeastern university in the US, found
high levels of strategy use for two-thirds of the strategy groups, including social,
metacognitive, cognitive, and compensation strategies, while medium levels of use were
uncovered for the other two strategy categories, affective and memory strategies. Phillips
(1990, 1991) investigated the strategies of 141 adult ESL learners in seven western states and
found that half of the strategy groups were used at a high level (metacognitive, social, and
compensation strategies), while the others were used at a medium level (cognitive, affective,
and memory strategies).

Chang (1991) used the SILL to investigate frequencies of strategy use of 50 Chinese ESL
students at a southeastern university in the US. (Additional measures, as discussed earlier,
were self-ratings of proficiency and two standardized proficiency tests.) In this study, Chang
found the highest use of compensation strategies and the lowest use of affective strategies.
Humanities and social science majors used more learning strategies than science majors.

Green's (1991, 1992) preliminary study of 213 students (not to be confused with Green and
Oxford's larger study of 374 students in 1993) at a Puerto Rican university showed that only
one strategy category was used at a high level: metacognitive strategies; the other categories
were used at a medium level: social, cognitive, compensation, affective, and memory
strategies. The island of Puerto Rico, though bombarded by English input through TV, radio
and movies, does not demand English as a survival tool; Puerto Rico is thus a hybrid of ESL
and EFL elements.
STRATEGY INVENTORY FOR LANGUAGE LEARNING 13

Oh (1992), who conducted a study involving 59 EFL students from the National Fisheries
University in Korea, found that the only strategy category used at a high frequency was
metacognitive strategies, while medium-frequency strategies were compensation, affective,
social, and cognitive, and memory strategies were used at a low frequency. The 332 Korean
university students in a different study by Park (1994) used all groups of strategies at a
medium frequency level. They used metacognitive, memory, and compensation strategies
somewhat more often than cognitive, social, and affective strategies, but all were in the
medium range. These students were all at the intermediate level of language proficiency on the
TOEFL.

In another investigation, Noguchi (1991) studied 174 junior high Japanese learners of EFL. In
the first author's reanalysis of the data using Noguchi's frequency codes, we found that almost
all the strategy groups had medium to low use. Social strategies were notably unpopular with
these Japanese junior high students, as were metacognitive and affective strategies; somewhat
more popular were memory and cognitive strategies.

Klassen (1994) conducted a study of learning strategies of 228 freshmen English students at
Feng-Chia University in Taiwan. Subscale frequency means were: compensation 3.36,
affective 2.72, metacognitive 2.86, social 2.72, cognitive 2.69, and memory 2.64. All these
frequencies represent medium use; compensation strategies edged toward high use but did not
reach it. Somewhat similar results were found by Yang (1994), who led a pre- and post-test
study involving 68 Taiwanese university students of English. Pre-test frequencies were all
medium (2.92-3.28) except for compensation strategies, which were slightly above medium
(3.57). Likewise, post-test frequencies were all medium (3.08-3.31), except for compensation
strategies (3.65), after a semester of discussing strategies.

The Puerto Rican, Korean, Japanese, and Taiwanese students in these studies (Green, 1991,
1992; Noguchi, 1991; Oh, 1992; Green and Oxford, 1993; Klassen, 1994; Park, 1994; Yang,
1994) did not need English for daily survival. They did not necessarily want to become expert
language learners and were not using most kinds of strategies often.

Frequency o f strategy use related to career interests


In contrast to many of the above results, in a study by Mullins (1992) of 110 Thai university
EFL students majoring in English, half of the strategy categories were used at a high level
(compensation, cognitive, and metacognitive), while the other half were used at a medium
level (social, affective, and memory strategies). The differences between the Green ESL/EFL
hybrid results, the Oh EFL results, and the Noguchi EFL results on the one hand, and the
Mullins EFL results on the other can be explained by the fact that Mullins' EFL subjects had
specialized career interests. Mullins' subjects were all majoring in English and were probably
self-chosen as better language learners. This speculation is supported by Indonesian EFL data
provided by Davis and Abas (1991), who found that 64 language faculty members from four
higher education institutions--all expected to be good language learners--showed high-
frequency use of five out of six strategy groups (metacognitive, social, compensation,
cognitive, and memory), with medium use of affective strategies.

Frequency o f strategy use related to the institution


Frequency of strategy use might sometimes be related to the prestige of the institution, which
14 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

is in turn related to the kinds of students who are accepted by the institution. Watanabe (1990)
investigated the strategies of 316 EFL students in Japan and used a principal components
analysis to create strategy categories. He found that one of his samples, which came from a
prestigious university, used language learning strategies more frequently than the other
sample, which came from a less prestigious college. The first sample used compensation and
affective strategies at a high level, with other strategy categories at a medium level. The
second sample used all strategy groups at a medium level.

Frequency of strategy use related to culture


Bedell (1993) presented 50-item SILL frequency data from a number of studies on a graph,
indicating low to high frequencies. The main message found in Bedall's graph is that different
cultural groups use particular kinds of strategies at different levels of frequency. Bedell's
study also included use of the 80-item SILL with 353 Chinese subjects and showed a linear
relationship between strategy use and proficiency.

Frequency of strategy use related to gender


In many ESL/EFL strategy frequency studies involving gender, the results have usually
favored females as more frequent users of strategies (for instance, Oxford et al., 1988,
1993a, b; Ehrman and Oxford, 1989; Green, 1991, 1992; Noguchi, 1991; Dreyer, 1992; Yang,
1992b, 1993; Green and Oxford, 1993; Oxford, 1993a, b). In a few studies, females have had a
distinctly different pattern of strategy use from that of males (Watanabe, 1990; Bedell, 1993).
Some studies, noted by Bedell and by Green and Oxford, have shown that males have
surpassed females on individual strategies but not on whole clusters or categories of strategies.

SEARCHING FOR THE UNDERLYING STRUCTURE OF THE ESL/EFL SILL:


RECENT FACTOR ANALYSES

A large meta-study (Oxford and Burry, 1993; Oxford, 1994) compared the factor structures of
six sets of ESL/EFL SILL data. The six sets included: (1) Green and Oxford (1993)--374
ESL/EFL ("hybrid") university students in Puerto Rico; (2) Yang (1992a)--590 university
EFL students in Taiwan, a larger group than Yang used in her dissertation (1992b, 1993); (3)
Zhang (1994)--741 secondary school EFL students in the People's Republic of China; (4)
Watanabe (1990)--255 university EFL students in Japan; (5) Boraie et al. (1994)--761 adult
EFL learners in Cairo, Egypt; and (6) the combination of Anderson (1993), Talbott (1993),
and Oxford et al. (1990)--137 university ESL students in the US.

A nine-factor, principal components, Varimax (oblique) solution was chosen for the six
studies, using the Kaiser rule of eigenvalues greater than 1.0. Factor loadings greater than or
equal to .30 were considered acceptable for simple structure. This solution accounted for over
50% of the variance in most data sets (51.6% in Puerto Rico, 51.9% in Taiwan, 43.7% in the
People's Republic of China, 53.3% in Japan, 44.4% in Egypt, and 51.9% in combined US).
This finding suggests that for most samples, about one half of the language learning strategy
use is represented by the items on the identified SILL factors. Bedell (1993) conducted exactly
the same kind of factor analysis with another group, 353 EFL students from the People's
Republic of China; however, Bedell used the 80-item version of the SILL rather than the
STRATEGY INVENTORY FOR LANGUAGE LEARNING 15

50-item ESL/EFL version, so the results, while broadly comparable, are not totally parallel.
Bedell's Factor 1 consisted of functional practice-productive strategies (active, naturalistic
l a n g u a g e use). B e d e l l ' s other factors included: (2) m e t a c o g n i t i v e ( m a n a g e m e n t ) ; (3)
compensation; (4) functional practice-receptive; (5) review and reception; (6) m e m o r y -
vocabulary; (7) formal practice and affective; (8) social and error-correction; (9) cognitive-
analytic. This overall structure accounted for 41% of the variance in SILL responses in the
Bedell study. Now we turn to the six factor analyses using the 50-item SILL (Table 1).

Table 1. Factor analysis comparisonacross six data sets

Factor/
Location PuertoRico Taiwan PR China Japan Egypt CombinedUS
1 Active Metacognitive Active Active Metacognitive Active
language use planning language use languageuse planning language use
2 Metacognitive Active Metacognitive Sensory Sensory Metacognitive
planning languageuse planning memory memory planning
strategies strategies
3 Affective Memory Affective Metacognitive/ Affective Affective
and social and analysis and social social/affective and social strategies
4 Reflection Formal Sensory Compensation Active Sensory
(analysis & oral practice memory and analysis language memory
anxiety) strategies use strategies
5 Sensory Social Compensation Formal Request Social
memory strategies in reading oral practice and strategies
strategies repetition
6 Social/ Compensation Metacognitive Affective Sensory Compensation
cognitive in reading and affective strategies memory and analysis
conversation and anxiety
7 Sensory Affective Sensory Compensation Compensation Metacognitive
(visual) strategies (visual) in speaking in reading planning
memory memory & listening
8 C o g n i t i v e Compensation Attention Attention General General
and in speaking to key to key memory memory
relaxation details details strategies strategies
9 General General General Reflection Sensory Compensation
compensation memory memory (analysis & memory and
strategies strategies anxiety) strategies nonanalytic

Puerto Rico
For Puerto Rico, Factor 1 was comprised of strategies for active, naturalistic language use
(21.6% of the variance explained by this factor). These strategies included reading for
pleasure, seeking opportunities to read in English, looking for people to talk to, writing notes
and other items in English, practicing English with others, reading without looking up all the
words, starting conversations in English, asking questions in English, watching TV or movies
in English, finding ways to use English, having clear goals, developing cultural understanding,
using new words in a sentence, and trying to talk like native speakers of English. Factor 2 was
primarily comprised of metacognitive planning strategies (7.1%), Factor 3 affective and social
16 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

strategies (4.8%), Factor 4 reflective strategies for language analysis and anxiety awareness
(3.8%), Factor 5 sensory memory strategies (3.7%), Factor 6 social and cognitive strategies for
conversation practice (3.1%), Factor 7 sensory, chiefly visual, memory strategies (2.6%),
Factor 8 strategies for cognitively manipulating the language and trying to relax (2.5%), and
Factor 9 general compensation strategies (2.4%).

Taiwan
For Taiwan, the most explanatory factor, Factor 1, was metacognitive strategies (26.3% of the
variance), including planning the schedule, looking for reading opportunities, having clear
goals, thinking about progress, reviewing often, making summaries, looking for conversation
partners, trying to find better ways to learn English, and noticing mistakes to learn better.
Factor 2 was clearly an active, naturalistic language use factor (4.9%), Factor 3 concerned
memory and analytic strategies (4.2%), Factor 4 formal oral practice strategies (3.8%), Factor
5 social strategies (3.3%), Factor 6 compensation in reading (2.9%), Factor 7 affective
strategies (2.6%), Factor 8 compensation in speaking (2.4%), and Factor 9 general memory
strategies (2.3%).

People's Republic of China


For the mainland Chinese sample in the Zhang study, Factor 1 consisted of active, naturalistic
language use and accounted for 18.9% of the variance. Strategies loading on this factor
included finding conversation partners, starting conversations in English, watching English
language TV or movies, finding ways to use English, writing notes and other items in English,
reading for pleasure, learning the culture, encouraging oneself to speak despite fear, trying to
talk like native speakers, guessing what the speaker will say, writing in a language learning
diary, using new words in sentences, using words in different ways, asking for correction,
looking for practice opportunities, and practicing with others. Factor 2 (4.7%) was devoted to
metacognitive planning strategies, Factor 3 (3.8%) affective and social strategies, Factor 4
(3.3%) sensory memory strategies, Factor 5 (2.9%) compensation in reading, Factor 6 (2.7%)
metacognitive and affective strategies, factor 7 (2.5%) sensory, mainly visual, memory
strategies, Factor 8 (2.5%) attention to key details, and Factor 9 (2.4%) general memory
strategies.

Japan
For Japan, strategies for active, naturalistic language use comprised Factor 1, the primary
explanatory factor (23.3% of the variance). These included starting conversations, looking for
people to talk with, writing in English, reading for pleasure, finding reading opportunities,
asking questions, using TV/radio, finding ways to use the language, learning about the culture,
encouraging oneself, avoiding translation, asking for help, using familiar words differently,
imitating speech, reading without looking up words, anticipating the speaker, finding a
different way to say something, and using gestures or the native language temporarily. Factor
2 chiefly concerned sensory memory strategies (6.7%), Factor 3 metacognitive, social, and
affective strategies (5.0%), Factor 4 compensation and analysis strategies (3.8%), Factor 5 for
formal oral practice (3.5%), Factor 6 affective strategies (3.3%), Factor 7 compensation in
speaking (2.8%), Factor 8 attention to key details (2.6%), and Factor 9 strategies for reflection
(related to analysis and anxiety) (2.5%).
STRATEGY INVENTORY FOR LANGUAGE LEARNING 17

Egypt
The Egyptian study of adult EFL learners produced the following factors. Factor 1, which
accounted for 10.7% of the variance, consisted mainly of metacognitive planning strategies
(finding out how to be a better language learner, thinking about progress, paying attention to
the speaker, finding as many ways as possible to use English, looking for people to talk to in
English, planning the schedule for studying, looking for opportunities to read in English)
supplemented by social, compensation, and active use strategies. Factor 2 (5.0%) was
composed of sensory memory strategies, Factor 3 (5.3%) affective and social strategies, Factor
4 (4.7%) active, naturalistic language use, Factor 5 (4.4%) request and repetition, Factor
6 (3.6%) sensory m e m o r y strategies and anxiety-reduction strategies, Factor 7 (4.4%)
compensation in reading and listening, Factor 8 (3.1%) general memory strategies, and Factor
9 (3.4%) sensory memory strategies.

Combined US
For the combined US sample, Factor 1 was made up of strategies for active, naturalistic
language use (16.6% of the variance), including imitating speech, using TV/radio, starting
conversations, using familiar words differently, practicing sounds or alphabet, finding ways to
use the language, guessing, encouraging oneself, writing in English, skimming, saying or
writing repeatedly, finding a different way to say something, trying to concentrate on the
speaker, planning goals, making summaries, putting new words into sentences, looking for
people to talk with, and reading for pleasure. Subsidiary factors included: Factor 2,
metacognitive planning strategies (8.5%); Factor 3, affective strategies (5.4%); Factor 4,
sensory memory strategies (5.2%); Factor 5, social strategies (4.1%); Factor 6, compensation
and analytic strategies (3.8%); Factor 7, metacognitive planning strategies (3.2%); Factor 8,
general m e m o r y strategies (3.2%); and Factor 9, compensation strategies not involving
analysis (3.0%).

Table 2. Common factors across data sets


Factor name Location and factor number
Active language use Puerto Rico 1, PR China 1, Japan 1, Combined US 1, Taiwan 2, Egypt 4
Metacognitive planning Taiwan 1, Egypt 1, Puerto Rico 2, PR China 2, Combined US 2 and 7
Sensory memory strategies Japan 2, Egypt 2 and 9, PR China 4, Combined US 4, Puerto Rico 5
Affective and social Puerto Rico 3, PR China 3, Egypt 3
Affective strategies Combined US 3, Japan 4, Taiwan 7
Reflection (analysis & anxiety) Puerto Rico 4, Japan 9
Formal oral practice Taiwan 4, Japan 5
Compensation and analysis Japan 4, Combined US 6
Compensation in speaking Japan 7, Taiwan 8
Social strategies Taiwan 5, Combined US 5
Sensory (visual) memory Puerto Rico 7, PR China 7
Attention to key details PR China 8, Japan 8
General memory strategies Combined US 8, Taiwan 9, PR China 9

Comparisons and comments


Among the most important factors explaining the variance were active, naturalistic language
use, metacognitive planning, and sensory memory strategies. These three factors appeared
repeatedly across data sets. Affective and social strategies as a combination, affective
18 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

strategies alone, reflective strategies, formal oral practice, compensation and analysis,
compensation in speaking, social strategies, visual memory, attention to key details, and
general memory strategies were also common to various data sets (Table 2).

Egypt stood out as the most unique data set (Table 3), with four factors not found elsewhere:
request and repetition, memory and anxiety, memory and compensation, and compensation
in reading and listening. Puerto Rico was also atypical, with three unique factors:
social/cognitive conversation, cognitive and relaxation, and general compensation. Taiwan
had two unique factors: first, memory and analysis; and second, compensation in reading. The
People's Republic of China, Japan, and combined US each had only one unique factor.
Because of the national and cultural differences in these factor analyses, it is obvious that the
same SILL factor structure might not be appropriate for all people who are learning ESL or
EFL. National/cultural differences exist, even though a particular individual might not fully
reflect the trends. It would be helpful in the future to create country-by-country SILL norms
around the world based on large-scale factor analyses. At this point we have only single data
sets from most countries with which to work. To make completely defensible national/cultural
norms, we would prefer several large data sets from each country.
Table 3. Unique factors in various data sets
Location Factor name and number
Puerto Rico Social/cognitive conversation (6), cognitive and relaxation (8), general compensation (9)
Taiwan Memory and analysis (3), compensation in reading (6)
PR China Metacognitive and affective (6)
Japan Metacognitive/social/affective (3)
Egypt Request and repetition (5), sensory memory and anxiety (6), compensation in reading and listening
(7), memory and compensation (8)
Combined US Compensation and nonanalytic (9)

IMPLICATIONS FOR RESEARCH AND INSTRUCTION

First, language researchers must conceptualize language learning strategies in a way that
includes the social and affective sides of learning (as shown in the SILL) as well as the more
intellectual and "executive-managerial" sides. Language learning, more than almost any other
discipline, is an adventure of the whole person, not just a cognitive or metacognitive exercise.

Second, through strategy assessment teachers can help their students recognize the power of
using language learning strategies for making learning quicker, easier, and more effective.
Teachers can help students identify their current learning strategies by means of surveys,
interviews, diaries, think-aloud protocols, or other means (for details see Cohen, 1987;
Oxford, 1990b). Teachers need to know the advantages and disadvantages of each assessment
technique. Multiple techniques are to be encouraged whenever the time and resources are
available. When time and resources are restricted, teachers should use the most reliable and
valid strategy assessment measure that they can; for many people to date, this has been the
SILL.

Third, based on the information from strategy assessment, teachers can weave learning
strategy training into regular classroom events in a natural, comfortable, but explicit way.
STRATEGYINVENTORYFOR LANGUAGELEARNING 19

C h a m o t and K u p p e r (1989), O x f o r d (1990b), and O ' M a l l e y and C h a m o t (1990) provide


helpful details on how to do this. Teachers must also keep in mind differences in motivation,
learning style, gender, and other factors that affect learning strategy use.

Fourth, strategy assessments using different measurement modes with the same sample of
students could be cross-correlated. This would contribute to the validity of various assessment
techniques. For instance, it would be useful to correlate results from a think-aloud protocol, an
interview, and a survey to see how closely they relate to each other. If results show that an
interview and a survey are highly correlated but that they are only weakly correlated with a
t h i n k - a l o u d p r o c e d u r e , this i n f o r m a t i o n w o u l d be useful in s e l e c t i o n of an a s s e s s m e n t
procedure.

Fifth, studies will need to be replicated so that more consistent information becomes available
within and across populations. Particularly important is more information on how students
from different cultural backgrounds and different countries use language learning strategies.
As shown above, students from different countries utilize different strategies and prioritize
common strategies differently. Further effort is underway to replicate this summary study and
develop norms for each specific country.

In sum, it is critical that learning strategies be considered when planning courses, teaching
students, and designing classroom research. Learning strategies germane to various countries
should be among the first considerations of any ESL/EFL teacher or researcher who wants to
enhance student learning.

Acknowledgement--Thanks to Martha Nyikos and Katalin Nyikos for field-testingthe earliest version of the ESL/EFL
SILL and for making suggestions about revising the instrument.

REFERENCES

ANDERSON, N. (1993) Data on adult ESL strategy use. Unpublished manuscript, Ohio University, Athens, OH.
BEDELL, D. (1993a) Chinese data on EFL strategy use among university students. Unpublished manuscript,
University of Alabama, Tuscaloosa, AL.
BEDELL, D. (1993b) Crosscultural variation in the choice of language learning strategies: A mainland Chinese
investigation with comparison to previous studies. Unpublished master's thesis, University of Alabama, Tuscaloosa,
AL.
BIALYSTOK, E. (1978) A theoretical model of second language learning. Language Learning 28, 69-83.
BIALYSTOK, E. (1981) The role of conscious strategies in second language proficiency. Modern Language Journal
65, 24-35.
BORAIE, D., KASSABGY, O. and OXFORD, R. (1994) Empowering teachers and learners: Style and strategy
awareness. Paper presented at the annual meeting of Teachers of English to Speakers of Other Languages, Baltimore,
MD.
CHAMOT, A. U. and KUPPER, L. (1989) Learning strategies in foreign language instruction. Foreign Language
Annals 22, 13-24.
CHAMOT, A. U., O'MALLEY, J. M., KUPPER L. and IMPINK-HERNANDEZ, M. (1987) A study of learning
strategies in foreign language instruction: first year report. InterAmerica Research Associates, Rosslyn, VA.
20 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

CHANG, S.-J. (1991) A study of language learning behaviors of Chinese students at the University of Georgia and the
relation of those behaviors to oral proficiency and other factors. Unpublished doctoral dissertation, University of
Georgia, Athens, GA.
COHEN, A. D. (1987) Studying learner strategies: How we get the Information. In Wenden, A. L. and Rubin, J. (eds),
Learner Strategies in Language Learning, pp. 31-42. Englewood Cliffs, NJ: Prentice-Hall.
CORRALES, O. and CALL, M. E. (1989) At a loss for words: The use of communication strategies to convey lexical
meaning. Foreign Language Annals 22, 227-240.
DAVIS, E. and ABAS, H. (1991) Second Language Learning Strategies Utilized by Some Members of Language
Departments at Four lnstitutions--Sulawesi, Indonesia. Sulawesi, Indonesia: Summer Institute of Linguistics.
DREYER, C. (1992) Learner variables as predictors of ESL proficiency. Unpublished doctoral dissertation,
Potchefstroom University, South Africa.
DREYER, C. and OXFORD, R. (1995) Learner variables related to ESL proficiency among Afrikaans-speakers in
South Africa. TESOL Quarterly (submitted).
EHRMAN, M. E. and OXFORD, R. L. (1989) Effects of sex differences, career choice, and psychological type on
adult language learning strategies. Modern Language Journal 73, 1-13.
EHRMAN, M. E. and OXFORD, R. L. (1990) Adult language learning styles and strategies in an intensive training
setting. Modern Language Journal 74, 311-327.
GREEN, J. (1991) Language learning strategies of Puerto Rican university students. Paper presented at the annual
meeting of Puerto Rico Teachers of English to Speakers of Other Languages, San Juan, PR.
GREEN, J. M. (1992) Additional analyses of Puerto Rican strategy data. Unpublished manuscript, University of
Puerto Rico at Mayaguez.
GREEN, J. and OXFORD, R. (1993) New analyses on expanded Puerto Rican strategy data. Unpublished manuscript,
University of Puerto Rico at Mayaguez and University of Alabama, Tuscaloosa, AL.
HARLOW, L. (1988) The effects of the yellow highlighter--Second-language learner strategies and their
effectiveness: a research update. Canadian Modern Language Review 45, 91-102.
HUANG, X.-H. (1984) An investigation of learning strategies in oral communication that Chinese EFL learners in
China employ. Unpublished master's thesis, Chinese University of Hong Kong.
HUANG, X.-H and VAN NAERSSEN, M. (1987) Learning strategies for oral communication. Applied Linguistics
8, 287-307.
KEEFE, J. W. and MONK, J. S., with LETTERI, C. A., LANGUIS, M. and DUNN, R. (1989). Learning Style Profile.
Reston, VA: National Association of Secondary School Principals.
K/M, J.-D. (1991) A comparison of learning strategies of college students enrolled in beginning and advanced English
as a second language. Unpublished doctoral dissertation, Loma Linda University, Loma Linda, CA.
KLASSEN, J. (1994) The language learning strategies of freshman English students in Taiwan: a case study. Master's
thesis, California State University at Chico, CA.
LAWRENCE, G. (1984) A synthesis of learning style research involving the MBTI. Journal of Psychological Type,
2-15.
McGROARTY, M. (1987) Pattems of persistent second language learners: elementary Spanish. Paper presented at the
annual meeting of Teachers of English to Speakers of Other Languages, Miami, FL.
MULLINS, P. (1992) Successful English language learning strategies of students enrolled in the Faculty of Arts,
Chulalongkorn University, Bangkok, Thailands. Unpublished doctoral dissertation, United States International
University, San Diego, CA.
MYERS, 1. and McCAULLEY, M. (1985) A Guide to the Development and Use of the Myers-Briggs Type Indicator.
Palo alto, CA: Consulting Psychologists Press.
STRATEGY INVENTORY FOR LANGUAGELEARNING 21

NOGUCHI, T. (1991a) Review of language learning strategy research and its implications. Unpublished bachelor's
thesis, Tottori University, Tottori, Japan.
NOGUCHI, T. (1991b). Questionnairefor Learners. Tottori University, Tottori, Japan.
NYIKOS, M. and OXFORD, R. (1993) A factor analytic study of language-learning strategy use: Interpretations from
information-processing and social psychology Modern Language Journal 77, 11-22.
OH, J. (1992) Learning strategies used by university EFL students in Korea. Language Teaching 1, 3-53.1.
O'MALLEY, J. M. and CHAMOT, A. U. (1990) Learning Strategies in Second Language Acquisition. Cambridge:
Cambridge University Press.
OXFORD, R. L. (1986) Development and psychometric testing of the Strategy Inventory for Language Learning
(SILL). ARI Technical Report 728. Alexandria, VA; Army Research Institute. Appendices as ARI Research Note
86-92.
OXFORD, R. L. (1986-present) Strategy Inventory for Language Learning. Various versions. Tuscaloosa, AL:
Oxford Associates.
OXFORD, R. L. (1989) Use of language learning strategies: A synthesis of studies with implications for strategy
training. System 17, 235-247.
OXFORD, R. L. (1990a) Language learning strategies and beyond: A look at strategies in the context of styles. In
Magnan, S. S. (ed.), Shifting the Instructional Focus to the Learner, pp. 35-55. Middlebury, VT: Northeast
Conference on the Teaching of Foreign Languages.
OXFORD, R. L. (1990b) Language Learning Strategies: What Ever), Teacher Should Know. Newbury House/Harper
& Row, New York. Now Boston: Heinle & Heinle.
OXFORD, R. L. (1993a) Instructional implications of gender differences in language learning styles and strategies.
Applied Language Learning 4, 65-94.
OXFORD, R. L. (1993b) La difference continue... : Gender differences in second/foreign language learning styles
and strategies. In J. Sutherland (ed.), Exploring gender, pp. 140-147. Englewood Cliffs, N J: Prentice-Hall.
OXFORD, R. L. and BURRY, J. A. (1993) Evolution, norming, and psychometric testing of Oxford's Strategy
Inventory for Language Learning (SILL). Paper presented at the annual meeting of the National Council on
Measurement in Education, Atlanta, GA.
OXFORD, R. L. and EHRMAN, M. E. (1989) Psychological type and adult language learning strategies: A pilot
study. Journal of Psychological Type 16, 22-32.
OXFORD, R. L. and EHRMAN, M. E. (1995) Adults' language learning strategies in an intensive foreign language
program in the United States. System in press.
OXFORD, R. L., EHRMAN, M. E. and LAVINE, R. Z. (1991) Style wars: Teacher-student style conflicts in the
language classroom. In S. S. Magnan (ed.), Challenges in the 1990sfor college foreign language programs, pp. 1-25.
Boston: Heinle & Heinle.
OXFORD, R. L. and NYIKOS, M. (1989) Variables affecting choice of language learning strategies by university
students. Modern Language Journal 73(2), 291-300.
OXFORD, R. L., NYIKOS, M. and EHRMAN, M. (1988) Vive la difference? Reflections on sex differences in use of
language learning strategies. Foreign Language Annals 21(4), 321-329.
OXFORD, R. L., NYIKOS, M., NYIKOS, K., LEZHNEV, V., EYRING, J. and ROSSI-LE, L. (1989) Learning
strategies of adult ESL learners in the US: A collaborative study. Unpublished manuscript, University of Alabama,
Tuscaloosa, AL.
OXFORD, R. L., PARK-OH, Y., ITO, S. and SUMRALL, M. (1993a) Factors affecting achievement in a satellite-
delivered Japanese language program. American Journal of Distance Education 7, 10-25.
OXFORD, R. L., PARK-OH, Y., ITO, S. and SUMRALL, M. (1993b) Learning a language by satellite: What
influences achievement? System 21(1), 314 8 .
22 REBECCA L. OXFORD and JUDITH A. BURRY-STOCK

OXFORD, R. L., TALBOTT, V. and HALLECK, G. (1990) Language learning strategies, attitudes, motivation, and
self-image of students in a university intensive ESL program. Paper presented at the annual meeting of International
Teachers of English to Speakers of Other Languages, San Francisco, CA.
PADRON, Y. N. and WAXMAN, H. C. (1988) The effects of ESL students' perceptions of their cognitive strategies
on reading achievement. TESOL Quarterly 22, 146-150.
PARK, G. (1994) Language learning strategies: why do adults need them? Manuscript, University of Texas at Austin.
PHILLIPS, V. (1990) English as a second language learner strategies of adult Asian students using the Strategy
Inventory for Language Learning. Unpublished doctoral dissertation, University of San Francisco, San Francisco, CA.
PHILLIPS, V. (1991, Nov.) A look at learner strategy use and ESL proficiency. CATESOL Journal 56-67.
POLITZER, R. (1983) An exploratory study of self-reported language learning behaviors and their relation to
achievement. Studies in Second Language Acquisition 6, 54-65.
POLITZER, R. and McGROARTY, M. (1985) An exploratory study of learning behaviors and their relationship to
gains in linguistic and communicative competence. TESOL Quarterly 19, 103-124.
ROSSI-LE, L. (1989) Perceptual learning style preferences and their relationship to language learning strategies in
adult students of English as a second language. Unpublished doctoral dissertation, Drake University, Des Moines, IA.
TALBOTT, V. (1993) Strategies of adult ESL students: Data. Unpublished manuscript, Skagit Valley Junior College,
Washington.
TAKEUCHI, O. (1991a) A study of EFL students' use of language learning strategies. Gohokenkyu to Eigokoyoiku
(Studies on English Language Usage and Language Teaching) 13, 58~55.
TAKEUCHI, O. (1991b) Language learning strategies in second and foreign language of strategies frequency.
Bulletin of the Institute for Interdisciplinary Studies of Culture 8, 64-83.
TAKEUCHI, O. (1993a) A study of language learning strategies and their relation to achievement in EFL listening
comprehension. Bulletin of the Institute of Interdisciplinary Studies of Culture 10, 131-141.
TAKEUCHI, O. (1993b) Language learning strategies and their relationship to achievement in English as a foreign
language. Language Laboratory 30, 17-34.
WANGSOTORN, A., SRIPAIPAN, T., RATTANAPRUCKS, N., JARUNGGIDANAN, T., SINGKALWANIJ, P.
and VEJAPHURTI, A. (1986) Relationships between learning modes and the beginners' success in English. Paper
presented at the annual meeting of International Teachers of English to Speakers of Other Languages, Anaheim, CA.
WATANABE, Y. (1990) External variables affecting language learning strategies of Japanese EFL learners: Effects
of entrance examination, years spent at college/university, and staying overseas. Unpublished master's thesis,
Lancaster University, Lancaster, UK.
WEINSTEIN, C. E., PALMER, D. and SCHULTE, A. C. (1987) Learning and Study Strategies Inventory.
Clearwater, FL: H & H Publishers.
WEN, Q. and JOHNSON, R. K. (1991) Language learning approaches and outcomes: A study of tertiary English
majors in China. Paper presented at the Sixth International Conference of the Institute of Language in Education,
Hong Kong, China.
WENDEN, A. and RUBIN, J. (eds) (1987) Learner Strategies in Language Learning. Englewood Cliffs, NJ: Prentice-
Hall.
WILDNER-BASSETT, M. (1992a) Data set of foreign language students' strategies. Unpublished manuscript,
University of Arizona, Tucson, AZ.
WILDNER-BASSET]?, M. (1992b) Relationships among language learning strategies, personality types, and learning
styles. Paper presented at the annual meeting of the American Council on the Teaching of Foreign Languages,
Chicago.
YANG, N.-D. (1992a) Data on students in an intensive EFL program in Taipei. Unpublished manuscript, National
Taiwan University.
STRATEGY INVENTORY FOR LANGUAGELEARNING 23

YANG, N.-D. (1992b) Second language learners' beliefs about language learning and their use of learning strategies:
A study of college students of English in Taiwan. Unpublished doctoral dissertation, University of Texas, Austin, TX.

YANG, N.-D. (1993) Understanding Chinese students' language beliefs and learning strategy use. Paper presented at
the annual meeting of International Teachers of English to Speakers of Other Languages, Atlanta, GA.
YANG, N.-D. (1994) An investigation of Taiwanese college students' use of English learning strategies. Research
report, National Taiwan University, Taiwan.
ZHANG, W. (1994) Data on university EFL language use. Unpublished manuscript, Ohio University, Athens, OH.

APPENDIX

STRATEGIES MEASURED ON THE SILL

(1) Think of relationships between known and new, (2) use new words in a sentence, (3) connect sounds and images,
(4) use mental images, (5) use rhyme, (6) use flashcards, (7) physically act out words, (8) review often, (9) remember
by location, (10) say or write words several times, (11) try to talk like native speakers, (12) practice sounds, (13) use
words in different ways, (14) start conversations, (15) watch TV/movies, (16) read for pleasure, (17) write notes,
messages, letters, or reports, (18) skim then read, (19) look for similar words across languages, (20) find patterns, (21)
divide words for meaning, (22) avoid verbatim translation, (23) make summaries, (24) guess the unknown, (25) using
gestures, (26) make up new words, (27) read without looking up words, (28) guess what the speaker will say, (29) use
circumlocution or synonym, (30) find as many ways as possible to use English, (31) notice mistakes, (32) pay
attention to the speaker, (33) find out how to learn better, (34) plan schedule, (35) look for conversation partners, (36)
look for opportunities to read, (37) have clear goals, (38) think about progress, (39) relax when fearful, (40) encourage
self to speak when afraid, (41) give self a reward, (42) notice tension, (43) write a learning diary, (44) talk about
feelings, (45) ask for slowness or repetition, (46) ask for correction, (47) practice with others, (48) ask questions, (50)
learn about culture.