Expanding the Motivation Construct in Language Learning

Author(s): Paul F. Tremblay and Robert C. Gardner

Source: The Modern Language Journal , Winter, 1995, Vol. 79, No. 4 (Winter, 1995), pp.
505-518
Published by: Wiley on behalf of the National Federation of Modern Language Teachers
Associations

Stable URL: [Link]

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@[Link].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
[Link]

National Federation of Modern Language Teachers Associations and Wiley are collaborating with
JSTOR to digitize, preserve and extend access to The Modern Language Journal

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 Expanding the Motivation Construct
in Language Learning
PAUL E TREMBLAY ROBERT C. GARDNER
The University of Western Ontario The University of Western Ontario
Department of Psychology Department of Psychology
London, Ontario London, Ontario
Canada, N6A 5C2 Canada, N6A 5C2
Email: tremblay@sscl. [Link] Email: gardner@sscl. uwo. ca

Recent reviews by Crookes & Schmidt (1991), D6rnyei (1994), and Oxford and Sh
(1994) have suggested that research concerned with motivation in second language ac
tion would benefit from a consideration of motivational constructs from other rese
areas. The present study addresses this issue by investigating the relation of a number o
measures of motivation such as persistence, attention, goal specificity, and causal attr
tions to each other, to existing measures of attitudes and motivation, and to indices
achievement in French courses. A sample of 75 students in a francophone secondary s
completed various motivational and attitudinal measures, and subsequently wrote a Fr
essay. Their final grades in the French course were later obtained from the school rec
Support was found for a LISREL structural equation model linking different aspects
motivation with language attitudes, French language dominance, and French achieve
It was concluded that the new motivational measures add to our understanding of mot
tion in language learning.

IN 1959, GARDNER AND LAMBERT FOUND primarily to Gardner's (1985) Socio-Educa-

that second language (L2) achievement was re-
tional Model, the authors recommended that
lated not only to language aptitude but alsoresearchers
to consider non-L2 approaches to mo-
motivation. Since then, several studies have tivation. Similar suggestions have since been
used different ways of assessing motivationalmade by Oxford and Shearin (1994) and D6r-
variables in various contexts and languages tonyei (1994). Although Crookes and Schmidt,
determine the role of motivation in learning an
Oxford and Shearin, and D6rnyei discuss a
L2 (e.g., Clement, Gardner, & Smythe, 1980; number of motivational theories and concepts
Clement & Kruidenier, 1985; D6rnyei, 1994; Ely,
from the psychological literature, none of them
1986; Gardner, 1985; Gardner, Lalonde, Moor-present empirical evidence that supports the va-
croft, & Evers, 1987; Gardner & MacIntyre, 1991;
lidity of these concepts as important compo-
Julkunen, 1989; Kraemer, 1993; Ramage, 1990).nents of L2 learning. In our opinion, empirical
The concept of language learning motivation
investigation is essential to demonstrate the
has become a central component of a numbertheoretical and pragmatic value of such integra-
of theories of L2 acquisition (e.g., Clement,
tions.
1980; Gardner, 1985; Krashen, 1981; Spolsky,The adoption of a wider vision of motivation
1985). presents an interesting challenge when we con-
In 1991, Crookes and Schmidt proposed that sider that there is no single motivational theory
research in L2 learning has been restricted by a suited to an understanding of all the factors
narrow perspective on motivation. Referring involved in motivational behavior (Landy &
Becker, 1987). Heckhausen (1991) captures the
vastness of what we call "motivation" when he
states:
The Modern Language Journal, 79, iv (1995)
0026-7902/95/505-520 $1.50/0
The observed goal-directedness of the behavior, th
?1995 The Modern Language Journal
inception and completion of a coherent behaviora

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 506 The Modern Language Journal 79 (1995)
unit, its resumption after an interruption, the tran-
tudes toward Learning the Language scales of
sition to a new behavioral sequence, the conflict
the Attitude/Motivation Test Battery (AMTB)
between various behavioral goals and its resolution,
(Gardner, 1985; Gardner, Clement, Smythe, &
all of these represent issues in motivation. (p. 9)
Smythe, 1979). Gardner (1985) proposes that all
The value of new motivational variables in a three components are necessary to describe
model could be judged on the basis of whether properly motivation in language learning. He
they provide any new theoretical insights. argues that effort by itself is not a complete de-
Cron-
bach and Meehl (1955) indicated in theirscription classic of motivation because individuals
paper on construct validity that learning mightmore expend considerable effort to please a
about a theoretical construct involves an elab- teacher or a parent without any real motivation
oration of the nomological network in which to learn the L2. Similarly, desire to learn the
the concept is found. A nomological network language
is or satisfaction with learning the lan-
best conceived as a diagram that outlines guagethedo not in themselves reflect true motiva-
connections among variables to explain ation. con-They must co-exist with effort. These lat-
cept or a process. One way of improving ter two
a components are somewhat related and
model is by clarifying the relationships can amongbe conceptualized as "valence" to refer to
its variables and reaching this objective the canvalue
be that an individual attaches to an out-
facilitated by the identification of mediators. A come.

mediator is a variable that links, and in a sense In the Socio-Educational Model (Gardner,
explains, the relationship between two other 1985) two classes of variables are hypothesized
variables. For example, the statement that diffi- to influence motivation. One of these is "Inte-
cult specific goals lead to increased perfor- grativeness," which refers to an open and posi-
mance does not answer the question of why tive regard for other groups and for groups that
such a relationship exists. However, goal-setting speak the language. It is currently assessed by
research suggests that setting difficult specific three measures from the AMTB: (a) Attitudes
goals leads to higher performance by activating toward the Target Language Group, (b) Inter-
variables such as effort. Thus this view suggests est in Foreign Languages, and (c) Integrative
Orientation. The second class of variables that
that goal setting directly influences effort (the
are seen in the Socio-Educational Model to in-
mediator), which in turn directly influences
fluence motivation is identified as Attitudes to-
performance.
In the present study, we address some of the ward the Learning Situation and is assessed by
above issues by investigating the relationships two AMTB measures: (a) Attitudes toward the
between motivational variables from Gardner's Language Course and (b) Attitudes toward the
(1985) Socio-Educational Model with measures Language Teacher. At a more general level, In-
of motivation derived from the psychological tegrativeness and Attitudes toward the Learn-
ing Situation have been categorized as Lan-
literature and measures of achievement in sec-
guage
ondary school French courses. We do this byAttitudes (Gardner et al., 1987).
determining how other measures of motivation,
derived from the general psychological litera-
EXPANDING THE MODEL
ture, would fit into an explicit expanded ver-
sion of the Socio-Educational Model. As a first step in presenting new motivatio
constructs, we make a distinction between mo-
tivational behavior and motivational anteced-
MOTIVATION IN THE SOCIO-
ents. Motivational behavior refers to the charac-
EDUCATIONAL MODEL
teristics of an individual that can be perceived
Gardner (1985) defines motivationby
toanlearn
observer. The variables that influence mo-
an L2 as "the extent to which the individual tivational behavior are often more difficult to
perceive by an outside observer but are self-
works or strives to learn the language because
reportable by the actor.1 In this paper, we will
of a desire to do so and the satisfaction experi-
refer to these latter variables as motivational
enced in this activity" (p. 10). This definition
includes three components: (a) effort ex- antecedents.

pended to achieve a goal, (b) a desire to learn

the language, and (c) satisfaction with the Variables
task Reflecting Motivational Behavior
of learning the language. These three compo-
nents can be assessed with the Motivational In- If we asked teachers to describe a motivated
student, their responses would likely include
tensity, Desire to Learn the Language, and Atti-

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 Paul E Tremblay and Robert C. Gardner 507

characteristics such as guided effort, by anticipatory goal states (Heck-

persistence, and
attention. These are meaningful concepts
hausen, 1991). Our cognitive ability tothat
antici-
lend themselves well aspate descriptors of motiva-
events or to form expectancies has impor-
tional behavior by an outside observer.
tant motivational Crookes
implications, because it also
and Schmidt (1991) suggest that
allows us to there
anticipate isrewards
and perceive an ex- that
tremely close link between motivation
follow a given behavior. The higherand at-
the expec-
tention. Kanfer and Ackerman (1989) indicate tancy that a behavior can produce a specific
that motivation refers to "the direction of atten- outcome, the greater tends to be the motivation
tional effort, the proportion of total attentional to perform the activity (Bandura, 1991). In the
effort directed to the task (intensity), and the context of language learning, one could hy-
extent to which attentional effort toward the pothesize that the language student may devote
task is maintained over time (persistence)" considerable effort and persistence if he or she
(p. 661). With respect to persistence, Maehr believes
and that his or her goal can be achieved.
Braskamp (1986) suggest that when a person Bandura (1989) suggested that the most im-
attends to a task for an extended period of expectancy that we learn is self-efficacy.
portant
time, observers are likely to infer that the indi-
Self-efficacy refers to an individual's beliefs that
vidual is motivated. Effort generally refershetoor she has the capability to reach a certain
the amount of energy expended (Maehrlevel & of performance or achievement. The im-
Braskamp, 1986) and is somewhat distinct fromplication of this construct is that the amount of
persistence. For example, an individual can
motivational behavior exerted to reach a spe-
spend much time at a task (persistence) but
cific outcome will be influenced by a perceived
work at a low rate of effort. Given such consid- probability of the attainability of the goal. In
eration it seems reasonable to assess motiva- the language learning literature, a construct
tional behavior using measures of effort, that
atten-
resembles expectancy and self-efficacy is
tion, and persistence. self-confidence, which is an important variable
Although such components are observable,
in Cl'ment's Social Context Model of L2 learn-
ing (Clement, 1980; Clement & Kruidenier,
they can also be assessed through self-reports,
1985). In that model, self-confidence involves a
and that strategy is used in this investigation.
The Motivational Intensity scale of the AMTB is anxiety when speaking an L2 and high
lack of
an appropriate measure of effort. Scales assess-
self-ratings of proficiency (see also Clement,
ing persistence and attention in French courses
Gardner, & Smythe, 1980). Clement and Kruid-
were developed for this study. enier (1985) proposed that in a multicultural
setting, self-confidence is the most important
Motivational Antecedents determinant of motivation to learn and use the
L2.
Motivational antecedents were defined as fac-
Self-confidence differs from self-efficacy
tors that cannot be readily perceived by an ex-
mainly in terms of the inclusion of an anxiety
ternal observer but still influence motivational
component. However, it may also be the case
behavior through their cognitive or affective in-
that low self-efficacy is accompanied by high
fluence. The reviews by Crookes and Schmidt
levels of anxiety. For example, an individual
(1991), Oxford and Shearin (1994), and D6rnyei
who has a high level of French Class Anxiety
(1994) of motivational research in the psycho-
may not expect to reach a high level of mastery.
logical literature discuss theories and concepts
Self-confidence in the language learning con-
such as intrinsic and extrinsic motivation, need
text is usually assessed with measures of per-
for achievement, expectancy-value, learned
ceived proficiency at the time of testing,
helplessness, and goal-oriented behavior. Many
of these theories refer to characteristics of the whereas self-efficacy is more closely tied to the
level of performance that an individual believes
individual that reflect motivation, and the pres-
he or she could achieve at some point in the
ent investigation will focus on a number of future.
them such as expectancy and self-efficacy, val-
In this study we will assess students' percep-
ence, causal attributions, and goal setting.
tions of their anticipated proficiency in French
Expectancy and Self-Efficacy. One concept that at the end of the course; this measure will be
stands out in current theories of motivation in labeled Performance Expectancy. In addition,
psychology is expectancy. Theories that focus French Use Anxiety and French Class Anxiety
on expectancy propose that organisms antici- from the AMTB will also be included. To the
pate events and that their behavior is thus extent that high levels of performance expec-

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 508 The Modern Language Journal 79 (1995)
tancy are strongly associated with
tionship low levels
to expectancy. of
Attributions that are as-
anxiety, this will suggest that anxiety
sociated with high could be could be labeled
self-efficacy
viewed as a debilitating component of self-
as adaptive attributions, whereas attributions
efficacy. associated with low levels of self-efficacy could
Valence. In the motivation literature, valence
be labeled as maladaptive attributions.
Based on the causal attribution literature,
is generally defined as the subjective value that
making attributions of success to ability would
an individual associates with a particular out-
be adaptive, whereas making attributions of
come (Lee, Locke, & Latham, 1989). Another
failure to lack of ability would be maladaptive.
way to define valence is simply in terms of desire Whereas attributions of success to effort and
and attractiveness toward the task. Heckhausen
failure to lack of effort would seem to be adap-
(1977) indicates that it is the consequence of
tive, research has shown inconsistent results.
the outcome that determines the level of incen-
Bandura (1991) explains this inconsistency by
tive value. This has important implications for
suggesting that although attributions of effort
language learning. As Oxford and Shearin
may seem to be related to self-efficacy, per-
(1994) indicated, if language students do not
ceived high levels of effort may at times imply
perceive value in their performance, then their
motivation will be lowered. It would seem that a low ability. Attributions to the context or to luck
in cases of success and failure would be mal-
necessary condition for perceiving value in
adaptive, because they represent external fac-
studying a language is an awareness of the tors
con-that the individual cannot control.
sequences (e.g., benefits) that would follow
By asking students whether they would attrib-
from the mastery.
ute their success to ability, effort, context, or
In the Socio-Educational Model (1985), the
luck, and their failure to lack of ability, lack of
scales, Desire to Learn French and Attitudes to-
effort, context, or bad luck, we can investigate
ward Learning French, along with Motivational
the relationship between their responses and
Intensity form the primary motivation con-
self-efficacy in terms of an adaptive-maladap-
struct. However, if we are attempting to sepa-
tive hypothesis.
rate motivation into its component parts, we
would suggest that the scales, Desire to Learn
Goal Setting. Goal setting theory (Locke &
French and Attitudes toward Learning French,
Latham, 1990) makes a clear prediction on how
measure a valence component of motivation.
to improve work performance. The theory sug-
Causal Attributions. Attribution theory is foun- gests that individuals who have accepted spe-
ded on the assumption that individuals seek to cific and difficult goals will outperform individ-
understand why events have occurred (Schus- uals with nonspecific ("do my best") and easy
ter, F6rsterlung, & Weiner, 1989). A postulate of goals. Lee, Locke, and Latham (1989) state that
this theory is that future behavior is in part de- the conceptualization of goals as immediate
termined by the perceived causes of past events regulators of behavior is central to goal setting
(Weiner, 1986). Internal attributions such as theory. In the context of language learning one
ability and effort are usually perceived as could hypothesize that individuals who assign
"within the individual," whereas external attri- themselves specific and difficult goals are bet-
butions such as luck and task difficulty are per- ter learners than individuals who do not have
ceived as "outside the individual." such goals. The hypothesis that goals regulate
Bandura (1991) suggested that ability attribu- effort expenditure has been supported by nu-
tions are associated with high self-efficacy. merous studies (see Locke & Latham, 1990).
Thus an attribution of success to ability is asso- More specifically, differences in effort have
ciated with high self-efficacy while an attribu- been found in studies that compare specific dif-
ficult goals with "do your best" goals. Studies
tion of failure to lack of ability is associated with
low self-efficacy. The literature on causal attri- have also shown that individuals with specific
butions in an achievement related context gen- and challenging goals persist longer at a task
erally suggests that the attributions have mo- than individuals with easy and vague goals.
tivational properties based on their influenceIn order to obtain an indication of whether
on expectancy (Weiner, 1986, 1992). difficult and specific goals are associated with
Although causal attributions can be cate- high levels of motivational behavior, we devel-
gorized in terms of whether they are internal oped
or two scales labeled Goal Specificity and
external to the individual, they may also be Goal dif- Frequency. Teachers would probably agree
ferentiated in terms of their hypothesized rela- that some students have no goals beyond that of

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 Paul F Tremblay and Robert C. Gardner 509
passing a course or obtaining a credit.
the bilingual [Link]
The hypotheses for this
students may have specific study
intentions
were derivedsuch as
in the context of a structural
equation
improving their speech so that modeling
they cananalysis.
give Figure
a 1 provides a
good presentation in front of the
proposed class. model
motivational Goal based on the pre-
ceding discussion
Specificity is designed to measure the extentand literature
to review. The
number goals
which students have specific associated with
in each arrow refers to a
their
French course; Goal Frequency is designed
specific hypothesis to
that is described below.
assess the extent to which students
Hypothesisset goalsthat
1 indicates forMotivational Behav-
themselves such as by making ior willplans.
have a direct
Theinfluence
two on Achievement.
This hypothesis
scales are related and are aspects of a more is supported
gen- by several studies
eral concept of Goal [Link] the Socio-Educational Model which show
that motivation influences achievement (e.g.,
CONTEXT OF THE STUDY Clement et al. 1980; Gardner, 1985; Gardner et
al., 1979; Gardner & MacIntyre, 1991).
Like some other studies concerned with the Hypothesis 2 indicates that French Language
Dominance
role of motivation in language learning (see, for will also have a direct influence on

example, Clement et al., 1980; Clement & achievement. French Language Dominance is
Kruidenier, 1985; Gardner, 1985), this study was mainly an indication of experience in French
conducted in a bilingual school environment. communication and, in this context, it is rea-
Subjects were registered in a French school, and sonable to hypothesize that such extracurricu-
the language under investigation was French. lar experience could influence achievement.
However, the language environment outside the Hypotheses 3, 4, and 5 state that Goal Salience,
school is mainly English, and although commu- Valence, and Self-Efficacy directly influence the
nication outside the classroom is vigorously en- level of Motivational Behavior respectively, and
couraged by the teachers to take place in French, Hypothesis 6 indicates that Adaptive Attributions
students frequently express themselves in Eng- have a direct influence on Self-Efficacy. These
lish. Also, some of the students use English in four hypotheses derive specifically from the lit-
their life environment more than French (see erature on motivation that was reviewed above.
subject description below). Regardless of the lan- Hypotheses 7, 8, and 9 state that Language Atti-
guage context, it seems reasonable to argue that tudes have direct influences on Goal Salience,
motivation will influence language achievement, Valence, and Self-Efficacy respectively. The pro-
though one might speculate on the role that posed relationship between Language Atti-
other attributes such as attitudes might play. tudes and Goal Salience is somewhat similar
In consideration of the nature of our sample, to that in Ajzen's theory of planned behavior
we developed a measure of French Language (1988) that indicates that attitudes toward a
Dominance to identify the role played by lan- behavior have direct influences on intentions.
guage background and experience. In addition, The proposed relationship between Language
we made use of grades in the French course as Attitudes and Valence is consistent with the
an index of language achievement, because this Socio-Educational Model. It will be recalled
was felt to be the most reflective of the goals of that, in the Socio-Educational Model, Inte-
the specific language experience under investi- grativeness and Attitudes toward the Learning
gation. We also, however, included an essay writ- Situation have a direct influence on motiva-
ing task to broaden the construct of language tion. In the present model, the direct influence
achievement in the model. Each of these meas- of Language Attitudes is on Valence. The pro-
ures, French Language Dominance, French posed relationship between Language Atti-
Grades, and Essay Writing will be discussed
tudes and Self-Efficacy is consistent with Clem-
more fully in the Method, Results, and Discus-
ent's Social Context Model (Clement, 1980;
sion sections.
Clement & Kruidenier, 1985) that suggests that
Language Attitudes influence self-confidence.
HYPOTHESES Hypothesis 10 states that French Language
Dominance has a direct influence on Adaptive
Given such considerations, we decided to Attributions. It will be recalled that attributing
base the motivational model on Gardner's one's success to ability is adaptive, while attrib-
(1985) Socio-Educational Model of L2 learning,
uting one's failures to lack of ability is maladap-
but to extend it to incorporate "new motiva-
tive. Because French Language Dominance is
tional variables," and to make it applicable
an to
indication of ability to use the French lan-

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 510 The Modern Language Journal 79 (1995)
FIGURE 1

A Proposed Motivational Model

GOAL
SALIENCE

7
3

LANGUAGE
ATTITUDES8/8 VALENCE -"- 3
4 MOTIVATIONAL
BEHAVIOR
5 1
SELF-EFFICACY

62 ACHIEVEMENT
ADAPTIVE
ATTRIBUTIONS

10

FRENCH
LANGUAGE
DOMINANCE

MOTIVATION

Note: Numbers refer to hypotheses presented in the text.

Scale,
guage, it seems reasonable to hypothesize (b) ait
that series of attitude and motivation
scale items presented in a random order, and
will be directly related to Adaptive Attributions.
(c) a performance expectancy measure. Third,
METHOD the French essays were written in class 3 weeks
after the administration of the questionnaire.
Subjects Final grades in the French course were ob-
tained from the school after the course was
The sample consisted of 75 students (31 males completed.
and 44 females; mean age = 17.8) enrolled in
French language courses in a Northern Ontario
francophone secondary school. The sample in- Measures
cluded 18 students in grade 11, 21 students in
grade 12, and 36 students in grade 13. Seventy- All questionnaire measures, with the excep-
six percent of the students reported that French tion of Performance Expectancy and French
was their first language, while 24% reported Language Dominance, made use of items that
that English was their first language. were rated on a 7-point Likert scale ranging
from strong disagreement (-3) to strong agree-
Procedure
ment (+3). The scales were administered in
French; the examples in the following descrip-
Testing occurred during May, 1993, one tions are in English. Responses on the nega-
month before course completion. Test material tively worded items were recoded before scor-
was administered in the following order. First, ing. Number of items and internal consistency
the students read a general instruction form reliability coefficients (Cronbach's Index of In-
and signed a consent sheet. Second, they an- ternal Consistency) obtained with this sample
swered the questionnaire that consisted of are presented in brackets along with scale de-
three parts: (a) a French Dominance Language scriptions and sample items.2

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 Paul F Tremblay and Robert C. Gardner 511

Attitude Motivation Testhension when called The

Battery. upon to use French. "I
following
11 scales are from the AMTB. All of these consist
would feel uncomfortable speaking French un-
der any circumstances."
of positively and negatively worded items ex-
cept the Integrative Orientation and Instru-
mental Orientation scales that consist of pos- French Class Anxiety (10 items; ao = .82). A high
score represents a considerable level of appre-
itively worded items only. An example of one
hension experienced when called upon to use
item for each scale is presented in quotes fol-
French in the French classroom. "I never feel
lowing the description of each scale.
quite sure of myself when I am speaking in our
Attitudes toward French Canadians (10 items; aFrench
= class."
.73). A high score represents a positive attitude
toward French Canadians. "I would like to know Attention and Persistence Measures
more French Canadians."
The following scales were developed by the
Interest in Foreign Languages (10 items; o =
authors for the present study and consist of five
.80). A high score represents an interest in
positively worded and five negatively worded
learning and using a language other items
thaneach.
French or English. "I often wish I could read
newspapers and magazines in another Attention (a = .84). A high score suggests a
language." high level of attention. "Nothing distracts me
when I am studying French."
Integrative Orientation (4 items; oa = .71). This
scale assesses the extent to which students seek Persistence (6 = .76). A high score suggests a
high level of persistence. "I work on my French
to learn French for integrative reasons. "Study-
ing French is important because it will enablehomework very regularly."
me to better understand French Canadian life
and culture." Causal Attribution Measures

Instrumental Orientation (4 items; a = .50). This

The following scales were adapted from the
measure assesses the degree to which students
Multidimensional-Multiattributional Causality
seek to learn French for pragmatic reasons.
Scales (Lefcourt, 1981; Lefcourt, Von Baeyer,
"Studying French is important because it will
Ware, & Cox, 1979) to refer specifically to a
give me an edge in competing with others."
language learning context. There are eight
Motivational Intensity (10 items; oa = .72). Ascales,
high with three items each, based on the four
score represents considerable effort expended
causal attributions of ability, effort, context,
and luck for both success and failure situations.
to study French. "I really work hard to learn
French."
Ability-Success (ox = .64). This scale assesses the
degree to which one ascribes success in French
Desire to Learn French (8 items; ox = .78). A high
score on this measure represents a strong desire
to ability. "The most important ingredient in
to study French.3 "I want to learn French so learning
well French is my overall language ability."
that it will become second nature to me."
Ability-Failure (ar = .68). This measure repre-
Attitudes toward Learning French (10 items; o sents
= the degree to which one ascribes failure in
.85). A high score represents a positive attitude
French to lack of ability. "If I did not succeed in
toward learning French. "I really enjoy learning
a French course, it would cause me to question
French."
my French ability."

Attitudes toward the French Teacher (10 items; aoEffort-Success

= (6L = .54). This measure assesses
.92). A high score represents a favourable atti- the degree to which one ascribes success in
tude toward the French teacher. "My French French to effort. "My success in learning French
teacher has a dynamic and interesting teaching is a direct result of my effort."
style."
Effort-Failure (ar = .71). This scale represents
Attitudes toward the French Course (10 items; c = the degree to which one ascribes failure in
.85). A high score represents a favourable atti- French to lack of effort. "When I receive a poor
tude toward the French course. "I look forward grade in a French course, it is because I haven't
to the time I spend in my French class." studied enough for that course."
French Use Anxiety (10 items; cx = .92). A high Context-Success (a = .26). This is a measure of
score represents a considerable level of appre- the degree to which one ascribes success in

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 512 The Modern Language Journal 79 (1995)
French to the context. "My stand accomplishments
the meaning of most Frenchin proverbs,"
French are due mainly to the "Speakquality
French wellof teach-
enough to be able to teach
ing." it at the secondary level," and "Work as a writer
for a French newspaper."
Context-Failure (ot = .64). This scale measures
the degree to which one ascribes failure in
French to the context. "When I receive a low French Language Dominance (oa = .90)
grade in my French course, it is because This
the scale assesses students' relative domi-
professor has failed to make the course nance
in- in French compared to English. The
teresting." measure consists of seven multiple choice item
Luck-Success (a = .62). This measure assesses reflecting perceived ability in French compar
to English and frequency of French usage.
the degree to which one ascribes success in
high score on this scale suggests that the stu
French to luck. "When I receive a high grade in
French, I believe that I have been fortunate." dent is dominant in French. Two sample item
from this scale are: "At home I: (a) always spe
Luck-Failure (a = .44). This measure represents French, (b) speak French most of the time, (
the degree to which one ascribes failure in speak French half the time, (d) rarely spe
French to bad luck. "My French grade would be French, (e) never speak French" and "If I had
higher if I had been more lucky during tests." to count from 1 to 10 in my head, I would: (
always count in French, (b) count in Frenc
most of the time, (c) count in French half the
Goal Salience
time, (d) rarely count in French, (e) never
count in French."
Two scales were developed by the authors to
assess the degree to which individuals have spe-
Achievement Measures
cific goals associated with studying French and
to determine the frequency of goal strategy use
French Essay. Specifically for this study, stu-
(e.g., making plans or schedules) to study dents wrote a one-page French essay on the
French. These scales consist of four positively
topic "My greatest fear." It should be noted that
worded and four negatively worded items each.
the essay did not contribute to the final grades.
Goal Specificity (oa = .72). A high score on thisThe essays were graded independently by a
scale suggests that an individual has specificgraduate student in French and by the senior
goals with respect to the French course. "I haveinvestigator using five items assessing grammar,
a clear idea of the level of French I want to quality of vocabulary, complexity of sentence
reach. " structure, content development, and content
originality.5 These were aggregated for each
Goal Frequency (oa = .75). A high score on this judge, and the two summed ratings were added
scale suggests frequent use of goal strategies to to form a composite score. The agreement in-
learn French. "I often make a list of the things I dex is .79, adjusted by the Spearman Brown for-
have to do in my French course." mula, indicating an acceptable level of re-
liability for the aggregate score.
French Course Grade. This measure was the
Performance Expectancy (oa = .91)
overall grade in the French course. The final
A performance expectancy scale was also cre- grade is based on tests of grammar and compre-
ated for this study based on the Can Do meas- hension of text, essay writing, and oral presen-
ure developed by Clark (1984). The total scale tations.
consists of 30 items from the categories of Un-
derstanding, Speaking, and Writing. Subjects Analysis: The Structural
were asked to indicate on a 7-point Likert scale Equation Modeling Approach
the likelihood that they could perform the spe-
cified French language activities by the end of One of the two basic parts involved in struc-
the course. The items in each category differ in tural equation modeling is the measurement
difficulty. A high score on this scale suggests model, which refers to the relationships be-
that an individual expects to reach an advanced tween latent variables and their corresponding
level of French by the end of the course.4 Sam- indicator variables. Essentially, a latent variable
ple items from each category include "Under- is an underlying phenomenon that the indica-

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 Paul E Tremblay and Robert C. Gardner 513
tor variables are intended to reflect. In our The Measurement Model
model, for example, Motivational Behavior is
the latent variable and motivational intensity
One can inspect the measurement properties
(effort), persistence, and attention are of
thethe present model by referring to the load-
indi-
ings presented
cator variables. The measurement properties of along the arrows leading from
a model can be investigated by inspectingthe latent
the variables (ellipses in this figure) to
magnitude of the "loadings" of each indicator
the indicator variables (boxes). For example, the
variable on its latent variable. Thesemeasures
loadings Motivational Intensity, Attention, and
Persistence
vary from -1 to 1 and reflect the correlation of have loadings of .85, .97, and .79,
the indicator variable with the latent variable. respectively, on the latent variable "Motiva-
tional Behavior." These loadings suggest that
To the extent that the loadings are relatively
high and statistically significant, there is
theevi-
three measures strongly reflect "Motiva-
dence that the corresponding indicatortional
vari-Behavior."
ables contribute to the definition of the latent Inspection of the components of the measure-
variable. ment model will indicate that all latent variables

The second part of structural equation mod- are significantly and appropriately defined by
eling is the structural model. This part involves their indicator variables. Slight modification had
the calculation of regression (or path) coeffi- to be made, however, to the latent variable
cients and determines whether the relation- "Adaptive Attributions." The Adaptive Attribu-
ships proposed by the paths are significant. In a
tions variable is represented by the attribution
diagram such as Figure 1, a path is represented measures of ability, context, and luck in success
by a line with an arrow. The direction and of failure
the situations. The effort attributions
arrow indicates the direction of influence. For were deleted from the model, because they did
example in Figure 1, there is a path leading not have statistically significant loadings. This
from Self-Efficacy to Motivational Behavior, lackin-
of significance is consistent with comments
dicating that Self-Efficacy influences Motiva- by Bandura (1991) who suggests that the relation-
tional Behavior. The advantage of structural ship between effort attributions and self-ef-
equation modeling over a simpler regression oris unclear. He argued that although effort
ficacy
correlational procedure is that it assesseswould rela- seem to be positively related to self-
tionships or paths simultaneously, thus reveal- efficacy, a high level of effort might also imply
ing unique relationships that are not low con-ability. Furthermore, there is evidence sug-
founded with other variables in the model. In gesting that children who attribute a failure to
addition to indicating whether paths are signifi- lack of ability become less motivated than chil-
cant, structural equation modeling can provide dren who attribute the failures to a lack of ef-
fit indices that indicate how well the model as a fort (Dweck & Leggett, 1988). The loadings on
whole fits the data. this dimension suggest that the Adaptive Attri-
butions variable is positively related to Ability-
RESULTS AND DISCUSSION Success, but negatively related to Ability Failure,
Context-Success, Context Failure, Luck-Success,
The model discussed in the introduction was and Luck-Failure.

A negative loading indicates that the in-

tested using the linear structural equation pro-
gram LISREL 7 (J6reskog & S6rbom, 1989). dicator
In variable in question has a negative as-
order to account for a possible confounding sociation with the latent variable. Thus, a high
score on Ability-Success is associated with a
factor of course level, we used a pooled within
course variance-covariance matrix based on the high score on Adaptive Attributions. High
three course levels as our data matrix.6 The fi-scores on the measures with negative loadings
nal LISREL model along with the standardized on the other hand are associated with a low
coefficients is presented in Figure 2, and all score
co- on Adaptive Attributions and in a sense
efficients are statistically significant. The final
can be viewed as maladaptive. Attributing one's
model differs from the initial model only with success to ability is adaptive, because it is as-
respect to the deletion of the effort attribu- sociated with high self-confidence. The op-
tions, the addition of one path, a correlation posite is true for lack of ability in failure sit-
between two latent variables, and three correla-
uations. The attributions of success and failure
tions between errors of measurement. These to luck and context are maladaptive, because
modifications are outlined below in the Struc- they are associated with a lack of volitional
tural Model section. control.

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 514 The Modern Language Journal 79 (1995)
FIGURE 2
The LISREL Model

GOAL .41 MOTIVATIONA

SALIENCE BEHAVOR

.67 .93 .93 .33.3518

SAFC VALENCE
.6

73[?? ATITUDES .29HIEVEME

S-.78 .65 63

86*L? .53 FR I ANGUAGE ATTIONS

1.00

.45 7 -.34 -.57 -.74 -.48

E8 E9 E10 E11 C12 -13

ABF Ability-Failure; ABS Ability-Success; AFC Attitudes toward French Cana

the French Course; AFT Attitudes toward the French Teacher; ALF Attitudes
ATT Attention; COF Context-Failure; COS Context-Success; DLF Desire to Learn French; FCA French
Class Anxiety; FLD French Language Dominance; FUA French Use Anxiety; GOF Goal Frequency; GOS
Goal Specificity; IFL Interest in Foreign Languages; IGO Integrative Orientation; ISO Instrumental
Orientation; LUF Luck-Failure; LUS Luck-Success; MIT Motivational Intensity; PER Persistence; PEX
Performance Expectancy

The Structural Model suggested that adding another path and allow-
ing a number of variables to correlate would
This section tests the validity of the 10 hy- improve the fit of the model to the data. First,
potheses proposed in the introduction. In Fig- allowing for a correlation between Language
ure 2, all regression coefficients are significant, Attitudes and French Language Dominance,
providing support for the hypotheses. For ex- represented by a double arrow, indicates that
ample, Hypothesis 1 stated that Motivational the two latent variables are indeed correlated
Behavior would have a significant effect on (r= .29). This suggests that individuals who are
Achievement. The standardized coefficient of dominant in French have more favorable Lan-
.35 for this path was significant and indicates
guage Attitudes than those dominant in Eng-
that a one unit increase in Motivational Behav- lish, and vice versa. A second modification to
ior would lead to a .35 unit increase in Achieve- the model was a direct path between French
ment. Note that this effect is independent of Language
the Dominance and Self-Efficacy. Thus,
effect of French Language Dominance on although French Language Dominance has an
Achievement (Hypothesis 2), which in this case indirect effect on Self-Efficacy through the me-
is represented by a standardized coefficient of
diation of Adaptive Attributions, the modifica-
.34. tion indices suggest that it also has a direct in-
In addition to the paths that were proposed fluence. The other modifications consisted of
as hypotheses, the LISREL modification indices allowances of correlations between the error

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 Paul E Tremblay and Robert C. Gardner 515
variances of Attitudes toward the French Course this relationship is similar to that found by
and Attitudes toward the French Teacher, Goal Clement (1980) and replicates findings in re-
Frequency and Motivational Intensity, French search concerned with his Social Context Model.
Use Anxiety and French Class Anxiety, and Atti- Clement and Kruidenier (1985) indicate that in
tudes toward Learning French and Perfor- multicultural settings "it is this self-confidence
mance Expectancy.7 which is the most important determinant of mo-
tivation to learn and use the second language"
The Model Fit (p. 24). Furthermore, the literature on expec-
tancy theory supports this relationship by sug-
The LISREL program provides a chi-squaregesting that high self-efficacy will lead to high
measure of the goodness of fit of the model to motivational levels (Bandura, 1991; Kirsch,
the data. The chi-square measure was 453.50 1986; Weiner, 1986).
with 260 degrees of freedom, p < .01. The chi- The model also specifies that Achievement is
square per degree freedom index was 1.74, directly influenced by French Language Domi-
which indicates an adequate fit in that it is wellnance and Motivational Behavior. It is clear that
below the maximum value of 5.0 suggested by dominance in a language, especially when there
Wheaton, Muthen, Alwin, and Summers (1977).are salient individual differences, would predict
Furthermore, two goodness of fit indices, the achievement scores. Even after French Lan-
Goodness of Fit Index and the Adjusted Good- guage Dominance has been accounted for how-
ness of Fit Index were moderate, .70 and .63 ever, Motivational Behavior is still a significant
respectively, which can be expected due to the determinant of Achievement.
complexity of the model.

Discussion of the Model LIMITATIONS, IMPLICATIONS, AND

FUTURE RESEARCH
The model proposed in this study suggests
that a number of variables mediate the relation- The structural equation modeling analys
ship between Language Attitudes and Motiva-confirmed the hypotheses concerning relati
tional Behavior. Three such mediators are Goal ships among the motivational constructs a
achievement. However, it must be noted that th
Salience, Valence, and Self-Efficacy. The results
suggest that specific goals and frequent refer- elegance of structural equation modeling a
ence to these goals lead to increased levelsmethod of of model development is accompanie
motivational behavior. These findings are con- by some limitations. First, it is the case tha
sistent with Lee et al.'s (1989) proposition that
model can never truly be confirmed. All tha
goals influence performance by increasing can in-be said is that the model fits or does not fit
tensity, persistence, and attention. The model the data, and this is determined by the ade-
also indicates that Goal Salience is influenced quacy of the measurement model and the struc-
by Language Attitudes. One reasonable expla- tural model as indicated by the estimated pa-
nation for this relationship is the idea that posi-
rameters, the chi-square value, and the goodness
tive language attitudes will orient students of tofit indices. The model, of course, should also
develop specific language learning goals. Stu- be consistent with findings from previous re-
dents who have negative attitudes are more search. In the case of the present study, these
likely not to give much consideration to what conditions were satisfied.
they would like to achieve in the French course. A second related problem involves the direc-
A second mediator of the attitude/motiva- tion of causation. When researchers conduct a
tional behavior relationship is Valence. Lan-
simple regression analysis such as one in which
guage Attitudes are shown to influence Val- motivation is a predictor and achievement is the
ence, and the causal path between Valence and criterion, they are in essence stating that mo-
Motivational Behavior suggests that higher tivation influences achievement. Structural
levels of Motivational Behavior result when equation modeling follows a similar approach
learning is valued. such that the researcher has to specify, a priori,
A third mediator of the attitude/motiva- the direction of the paths, but it is always possi-
tional relationship is Self-Efficacy. In the pres- ble that other paths, including reciprocal paths,
ent model, Self-Efficacy is shown to be influ- among variables would improve the model. LIS-
enced by Language Attitudes and, in turn, to REL permits the researcher to determine if ad-
influence Motivational Behavior. The nature of ditional paths will improve the model, and in

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 516 The Modern Language Journal 79 (1995)
manipulation
the present case, no further actually improves
improvement wasmotivational
indicated. behavior.

A third concern is the size of the sample.

Improving self-efficacy is probably not as
Clearly, the larger the sample, the more stable
straightforward as improving goal setting. How-
the results would be across similar [Link] other research, in addition to the present
Small samples sometimes are unable to produce
study, has suggested that some causal attribu-
a solution (referred to as nonconvergence),tions
and have an influence on self-efficacy. Attri-
this is most likely to occur when the relation-
bution retraining techniques have revealed en-
couraging findings (F6rsterling, 1985). These
ships among the variables are not very strong.
Generally the larger the relationships amongare techniques in which behaviors that are con-
the variables, the more likely LISREL will sidered
con- to be maladaptive (e.g., attributing fail-
verge. Although our analysis was performed on to low ability) are altered by teaching more
ures
a sample of 75 subjects (somewhat less thanfavourable
the causal attributions (e.g., ascriptions
minimum size of 100 recommended by Boomsma, of failure to insufficient effort; see F6rsterling,
1982), we did not encounter convergence prob-
1985). After reviewing 15 attributional training
lems. Moreover, there are many parts of our studies, F6rsterling concluded that the methods
model that have been investigated in other have been consistently successful in increasing
persistence and performance. Such procedures
studies (e.g., Clement & Kruidenier, 1985; Gard-
ner, 1985; Kraemer, 1993), and those aspectsshould be investigated in the language learning
replicate here. We are confident, therefore,
context.

that the basic relationships reported here are

Self-efficacy in this study includes an anx
relatively stable. component, such that high anxiety levels
In consideration of the above limitations, we associated with low performance expecta
stress that our model is not carved in stone but
evaluations. The French Use Anxiety and Fre
Class
provides initial support for the development of Anxiety measures are conceived of as
more elaborate motivational theories of L2 tively stable characteristics of individuals.
learning. The new motivational variables is con-
not to say that intervention to reduce anx
tribute to a comprehensive view of the levels
motiva-in the language class is not possibl
tional process based on current motivational
fact, Horwitz and Young (1991) present ma
theories. Another benefit of this extension is a ways in which anxiety can be reduced in
generation of pragmatic hypotheses for influ- language learning environment.
encing motivational behavior. The model sug- Although the attitudinal and motivatio
gests that goal setting, valence, and self-efficacymeasures in this study are also conceptuali
have an influence on the level of motivational as relatively stable measures, they too are
behavior. Thus, when a teacher observes that a ject to modification as suggested by the So
student displays high levels of motivational be- Educational Model (Gardner, 1985). Charac-
havior, he or she may hypothesize that this con- teristics of the classroom that are under a
duct is influenced by the fact that the student teacher's control, for example, can interact w
sets goals for himself or herself, values the lan- an individual's motivational dispositions to p
guage course, and has a high level of self- duce motivational behavior that is manifested
efficacy. in terms of effort, attention, and persistence.
Future studies could investigate this hypothesis.
To what extent can one alter goal setting, val-
ence, self-efficacy, causal attributions, languageIn summary, this study is a first attempt to
attitudes, and motivational behavior? Future investigate the role that different aspects of mo-
language studies should address these ques-tivation play in the language learning context.
The findings also suggest that the measures in
tions by investigating characteristics of the en-
vironment that lead to changes in either ofthe Attitude/Motivation Test Battery (Gardner,
1985) are applicable to a bilingual context. It is
these attributes. For example, one could test the
hoped that this study, as one anonymous re-
validity of goal setting manipulations in the
classroom. Students could be asked to write es-
viewer stated, "will raise further important
says outlining their goals in the course and questions
the about the relationships among mo-
tivational variables and may well spark addi-
value of these goals. Throughout the course,
students could be given some feedback tional
and research to replicate the study or further
test of the model." We support Cronbach and
could be asked to evaluate their progress rela-
Meehl's
tive to their goals. Given our present model, it (1955) view that, in order to learn more
would be meaningful to determine whether about
this a theoretical construct, it is necessary to

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UT6 [Link] UTC
All use subject to [Link]
 Paul E Tremblay and Robert C. Gardner 517
elaborate on the network of variables associated at a point in time, whether a goal to reac
with that construct. We have employed such present
a or not.
5 Each item correlated above .85 with the total rat-
strategy here, showing how other elements of
motivation can be incorporated into the Socio- ing scale suggesting that all five items are highly
related.
Educational Model (Gardner, 1985) of L2 learn-
6 Although several structural equation modeling
ing. We believe the elaboration is fruitful andstudies continue to use a correlation matrix, Cudeck
conducive to further understanding and re-
(1989) discusses possible complex problems associ-
search. We do not believe, however, that this ated with the use of this type of matrix and suggests
calls for any reconceptualization of Gardner's using a variance-covariance matrix.
Socio-Educational Model. 7 Each indicator variable has an associated unique
Future research on the relationships between
variance term represented by a delta or epsilon sym-
motivational dispositions, motivational bol. These elements represent the variance in the in-
aspects
of the classroom, and motivational dicator
changes variables that is not explained by the latent
variables.
would not only extend our theoretical knowl-
edge of language learning motivation, but
would also suggest ways to improve motivation.
REFERENCES

ACKNOWLEDGMENTS

Ajzen, cago:
I. (1988). Attitudes,
Dorsey Press. personality, and behavior. Chi-
The preparation of this article was facilitated by a
Bandura, A. (1989). Self-regulation of motivation and
doctoral fellowship (752-94-1116) to the first author
and by a grant (410-93-0076) to the second author action through internal standards and goal sys-
tems. In A. Pervin (Ed.), Goal concepts in person-
for research on the topic "Motivation, Attitudes, Anx-
ality and social psychology (pp. 19-85). Hillsdale,
iety and Second Language Acquisition" from the So-
cial Sciences and Humanities Research Council of NJ: Lawrence Erlbaum Associates.
Canada. Bandura, A. (1991). Self-regulation of motivation
through anticipatory and self-reactive mecha-
We express our appreciation to H. Fournier and to
nisms. In R. A. Dienstbier (Ed.), Nebraska Sympo-
the other teachers for their effort during the data
sium on Motivation 1990:Perspectives on motivation
collection. We also thank V. Galbraith for her help
during the preparation of the data file and Z. (pp. 69-164). Lincoln: University of Nebraska
Press.
Lafargue for her assistance in grading the essays.
Boomsma, A. (1982). The robustness of LISREL
against small sample size in factor analysis
models. In K. G. J6reskog & H. Wold (Eds.),
NOTES
Systems under direct observation: Causality, structure,
prediction (Pt. 1) (pp. 149-173). Amsterdam:
North-Holland.
1 Although self-reports may be influenced Clark, J. L. D. (1984). Language. In T. S. Barrows
by social
desirability, some research has indicated that for meas-
(Ed.), A survey of global understanding: Final report
ures of the Socio-Educational Model, this influence(pp. 87-100). New Rochelle, NY: Change Maga-
is minimal (Gardner & Gliksman, 1982; Gardner zine &
Press.
MacIntyre, 1991). Furthermore the effects of social
Clement, R. (1980). Ethnicity, contact and communi-
desirability are relatively small in a context where
cative competence in a second language. In
there is no strong reason for participants to present
H. Giles, W. P. Robinson, & P. Smith (Eds.), Lan-
themselves in a favourable way. guage: Social psychological perspectives (pp. 147-154).
2 All measures had adequate reliability coefficients
Oxford, England: Pergamon Press.
(above .70) except a few of the attribution Cl6ment,
scales and R., Gardner, R. C., & Smythe, P. C. (1980).
Instrumental Orientation. In estimating the fit of and
Social a individual factors in second lan-
model, the structural equation modeling approach guage acquisition. Canadian Journal ofBehavioural
takes into account the level of reliability of the 12, 293-302.
Science,
measures.
Cl6ment, R., & Kruidenier, B. G. (1985). Aptitude,
3 Two items were deleted from the Desire attitude to
andLearn
motivation in second language
French scale since they were judged to be inappropri-
proficiency: A test of Cl6ment's model. Journal
ate in the present language context. These two
of Language and items
Social Psychology, 4, 21-37.
were: (a) "I wish I had begun studying
Cronbach,French at an
L. J., & Meehl. P. E. (1955). Construct va-
early age," and (b) "I wish I were fluent in
lidity in French."
psychological tests. Psychological Bulletin,
52, 281-302.
4 Expectancy in this case does not necessarily imply
a goal; it is simply a self-rating of level
Crookes,of
G., performance
& Schmidt, R. W. (1991). Motivation: Re-

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]
 518 The Modern Language Journal 79 (1995)
opening the research agenda.
[Link] Learn-
Journal of Applied Psychology Monograph, 74,
ing, 41, 469-512. 657-690.
Cudeck, R. (1989). Analysis Kirsch,
of correlation matrices
I. (1986). Early research on self-efficacy: What
using covariance structure models. Psychological
we already know without knowing we [Link]-
Bulletin, 105, 317-327. nal of Social and Clinical Psychology, 4, 339-358.
Kraemer, motivation
D6rnyei, Z. (1990). Conceptualizing R. (1993). Social psychological
in for-factors re-
lated to the
eign language learning. Language study of Arabic
Learning, among Israeli high
40,
45-78. school students. Studies in Second Language Acqui-
D6rnyei, Z. (1994). Motivation and motivating in the sition, 15, 83-105.
foreign language classroom. Modern Language Krashen, S. D. (1981). Second language acquisition and
Journal, 78, 273-284. second language learning. New York: Pergamon.
Dweck, C. S., & Leggett, E. (1988). A social-cognitive
Landy, E J., & Becker, W. S. (1987). Motivation theory
approach to motivation and personality. Psycho- reconsidered. Research in Organizational Behavior;
logical Review, 95, 256-273. 9, 1-38.
Ely, C. M. (1986). Language learning motivation:Lee,
A T. W., Locke, E. A., & Latham. G. P. (1989). Goal
descriptive and causal analysis. Modern Language setting theory and job performance. In A. Per-
Journal, 70, 28-35. vin (Ed.), Goal concepts in personality and social psy-
F6rsterling, E (1985). Attributional retraining: A re- chology (pp. 291-326). Hillsdale, NJ: Lawrence
view. Psychological Bulletin, 98, 495-512. Erlbaum.

Lefcourt, H. M. (1981). The construction and devel-

Gardner, R. C. (1985). Social psychology and second lan-
guage learning: The role of attitudes and motivation. opment of the Multidimensional-Multiattribu-
London: Edward Arnold. tional Causality Scales. In H. M. Lefcourt (Ed.).
Gardner, R. C., Cl6ment, R., Smythe, P. C., & Smythe, Research with the locus of control construct. Vol. 1. As-
C.L. (1979). Attitudes and Motivation Test sessment methods (pp. 245-277). New York: Aca-
Battery-Revised manual. (Research Bulletin demic Press.

#15). London, ONT: University of Western Lefcourt, H. M., Von Baeyer, C. L., Ware, E. E., & Cox,
Ontario. D. J. (1979). The Multidimensional-Multiattri-
Gardner, R. C., & Gliksman, L. (1982). On 'Gardner butional Causality Scale: The development of a
on affect': A discussion of validity as it relates to goal specific locus of control scale. Canadian
the Attitude/Motivation Test Battery: A re- Journal of Behavioural Science, 11, 286-304.
sponse from Gardner. Language Learning,Locke,
32, E. A., & Latham, G. P. (1990). A theory of goal
191-200. setting & task performance. Englewood Cliffs, NJ:
Gardner, R. C., Lalonde, R. N., Moorcroft, R., & Prentice Hall.

Evers, F T. (1987). Second language attrition: Maehr, M. L., & Braskamp, L. A. (1986). The motivation
The role of motivation and use. Journal of Lan- factor A theory of personal investment. Lexington,
guage and Social Psychology, 6, 29-47. MA: Lexington Books.
Gardner, R. C., & Lambert, W. E. (1959). Motivational Oxford, R., & Shearin, J. (1994). Language learning
variables in second language acquisition. Cana- motivation: Expanding the theoretical frame-
dian Journal of Psychology, 13, 266-272. work. Modern Language Journal, 78, 12-28.
Gardner, R. C., & MacIntyre, P. D. (1991). An instru- Ramage, K. (1990). Motivational factors and persis-
mental motivation in language study: Who says tence in foreign language study. LanguageLearn-
it isn't effective? Studies in Second Language Acqui- ing, 40, 189-219.
sition, 13, 57-72. Schuster, B., F6rsterlung, E, & Weiner, B. (1989). Per-
Heckhausen, H. (1977). Achievement motivation and ceiving the causes of success and failure. A
its constructs: A cognitive model. Motivation and cross-cultural examination of attributional
Emotion, 1, 283-329. concepts. Journal of Cross-Cultural Psychology, 20,
Heckhausen, H. (1991). Motivation and action. Berlin: 191-213.

Springer-Verlag. Spolsky, B. (1985). Formulating a theory of second

Horwitz, E. K., & Young, D. (1991). Language anxiety: language learning. Studies in Second Language Ac-
From theory and research to classroom implications. quisition, 7, 269-288.
Englewood Cliffs, NJ: Prentice Hall. Weiner, B. (1986). An attributional theory of motivation and
J6reskog, K. G., & S6rbom, D. (1989). LISREL 7: A emotion. New York: Springer-Verlag.
guide to the program and applications. Chicago: Weiner, B. (1992). Human motivation: Metaphors, theories,
SPSS. and research. Thousand Oaks, CA: Sage Publica-
tions.
Julkunen, K. (1989). Situation- and task-specific motivation
Wheaton, B., Muthen, B., Alwin, D. F., & Summers,
in foreign language learning and teaching. Joensuu,
Finland: University of Joensuu, Publications in G. E (1977). Assessing reliability and stability in
Education. panel models. In D. R. Heise (Ed.), Sociological
Kanfer, R., & Ackerman, P. L. (1989). Motivation and Methodology (pp. 84-136). San Francisco: Jossey-
Bass.
cognitive abilities: An integrative/aptitude-
treatment interaction approach to skill acquisi-

This content downloaded from

[Link] on Fri, 01 Jul 2022 [Link] UTC
All use subject to [Link]