Professional Documents
Culture Documents
Social-Psychological Theory of
Second-Language (L2) Learning*
Shun Y. Au**
City University (London)
GARDNER’S THEORY
SOCIALPSYCHOLOGICAL
OF L2 LEARNING: AN OVERVIEW
75
76 Language Learning Vol. 38, No. I
LINGUISTIC
LEARNING
SITUATION
NONLlNGUlSTlC
with the task of not simply learning new information (vocabulary, gram-
mar, pronunciation, etc.) which is part of his own culture but rather of
acquiring symbolic elements of a different ethnolinguistic community.”
(1979, p. 193). If indeed it is accepted that L2 learning is primarily a
social-psychological phenomenon, it is only natural that social-
psychological variables should be given paramount attention. With this
in mind, Gardner and his associates have been continuously working
with a set of social-psychological variables over the past 25 years. These
variables have now been incorporated into a theoretical model, the sche-
matic form of which is presented in Figure 1. As can be seen from Figure
1, the model has four major sections. They are:
1. social milieu,
2. individual differences
3. language acquisition contexts, and
4. outcomes
that the cultural beliefs within this milieu could influence the development
of two sets of attitudinal variables relevant to language acquisition. These
two attitude variables are Integrativeness (which refers to positive affect
towards the other language community, or communities in general) and
Attitudes toward the Learning Situation (evaluative reactions to the learn-
ing situation). It is hypothesised that these two attitudes influence motiva-
tion, and the composite of all three classes of variables is referred to as an
integrative motive. Motivation in this context, it should be emphasised, is
conceived as comprising three components, effort to achieve the goal,
desire to achieve the goal and positive affect toward the goal. Such a tri-
partite conceptualisation seems necessary to reflect adequately the notion
of the motivated individual.
The model proposes further that two individual difference variables,
motivation and language aptitude, interact with second language acquisi-
tion contexts to promote second language proficiency. Both aptitude and
motivation are expected to play a role in formal language acquisition
contexts, like the school classroom, because the student with more ability
and a higher level of motivation would be expected to work harder and
profit more than the student with lower levels. Motivation is expected to
play a more dominant role than aptitude in informal contexts, such as
watching movies in the other language, because it is assumed that motiva-
tional levels will determine whether or not students avail themselves of
such informal opportunities to learn the language. Once the student enters
an informal language acquisition context, language aptitude would become
influential, but motivational factors dominate because they determine
whether or not students even take advantage of these contexts. Finally, the
model shows that both contexts give rise to both linguistic and non-
linguistic outcomes. Linguistic outcomes refer simply to language knowl-
edge and language skills, whereas non-linguistic outcomes refer to interest
in learning more of the language, desire to use it, etc. (pp. 222-223)
All five propositions have been explicitly stated by Gardner and his associ-
ates and directly or indirectly tested by them or by other researchers.
Before we discuss these propositions, however, it seems necessary to look
at the operational measures used to elicit one major construct: the integra-
tive motive.
The first four scales form the integrativeness component, the next three
the motivation component, the eighth and the ninth the attitudes toward
the learning situation component of the model. It should be noted that
this taxonomy for combining scales into components is not of an empiri-
cal nature in that Gardner and his associates have not employed a factor
analytic (or other) statistical technique to justify this taxonomy. Gardner
himself freely admits this classification system is a logical rather than an
empirical one (Gardner, 1980). Scores of all the scales are added together
to arrive at a single score of the integrative motive. It is not at all clear on
which grounds such a linear combination is justified. If it is proposed
that the nine scales constitute three separate components, and yet scores
of the three components can be additionally combined to arrive at one
single score, one can only conclude that either the subdividing of the nine
scales into three separate components is no more than an empty rhetoric
or the basis for adding the nine scales is an insecure one.
The scales have been administered to 5,000 English Canadian pupils,
grades 7 to 1 1, in seven regions of Canada yielding reasonably high inter-
nal consistency estimates and moderate test-retest reliability estimates.
It should be noted that the nine scales have seldom been employed in
their entirety in one study. A typical study by Gardner and his colleagues
would use some scales from the AMTB, some other variables thought to
be important for the particular sample under study, and some measures
of L2 achievement. Moreover, researchers other than Gardner and his
colleagues have sometimes adapted scales from the AMTB for use in
their own studies (Muchnich & Wolfe, 1982; Wong, 1982), and at other
times have devised entirely new scales (Spolsky, 1969; Lyczak, Fu, & Ho,
1976; Oller, Baca, 8z Vigil, 1977; Oller, Hudson, & Liu, 1977). Because
estimates of reliability and validity were seldom given in these studies,
when disconfirming evidence was found, it is difficult to judge to what
extent it was a result of inappropriate measures, and to what extent it
reflected the falsehood of the theory.
with the Spolsky identity scales, this method is a more global measure of
attitude toward linguistic communities, and therefore provides no test of
Gardner’s specific proposals for defining integrative motive. The extent
to which it adequately reflects the integrative motive is therefore
unclear.
With this knowledge of the various operational measures most fre-
quently employed in mind, we move on to evaluate the five major propo-
sitions of Gardner’s theory.
Wolfe (1982), Wong (1982), and Au (1984) provided strong evidence for
a lack of generality because they employed scales adapted from the
AMTB that were subsequently found to have high reliability with the
subjects in question, and yet no relationship or a negative relationship
between scores on them and L2 achievement were observed.
PROPOSITION 3: THEACTIVELEARNER
HYPOTHESIS
Proposition 3 maintains that integratively motivated L2 learners
achieve high L2 proficiency because they are active learners. Studies
relevant to this proposition have sought to prove and have indeed found
that subjects with high integrative motive tended to be more active in L2
learning situations, whether they be formal learning contexts (Gliksman,
Gardner, & Smythe, 1982) or informal learning contexts (Cltment,
Gardner, & Smythe, 1975). Moreover, those who had low integrative
motive were more likely to drop out of the L2 learning program (Bartley,
1969, 1970; Gardner & Smythe, 1975). However, all these studies suf-
86 Language Learning Vol. 38, No. 1
RECENTDEVELOPMENTS
Recent developments in Gardner’s theory have involved (1) explicating
and empirically investigating the role integrative motive plays in bringing
about better L2 achievement; (2) asserting the causal role of the integra-
tive motive by empirical investigation using the causal modeling method;
and (3) postulating a further motivational construct -self-confidence in
L2 - and relating it to the theory by claiming that this proposed construct
Au 89
The basis for ClCment’s proposal is his own research conducted within
the framework of the social-psychological model proposed by Gardner.
Two studies cited above (ClCment, Gardner, & Smythe, 1977, 1980) seem
to suggest that although some variables of the integrative motive are
related to some measures of L2 achievement, the majority of the integra-
tive motive variables does not seem to be associated with L2 achieve-
ment. Furthermore, the relationship in these studies is rather modest
compared to that found between L2 achievement and self-confidence
measures (anxiety associated with L2 learning and self-ratings of
proficiency- regarded as reflecting one’s confidence in one’s L2). From
these two studies, ClCment (1980) proposed not only that self-confidence
in L2 may form the basis of a secondary motivational process, but also
that it could at times assume a more important role than the integrative
motive in the learning of a L2.
One common weakness of the (ClCment et al.) studies cited above
seems to be the operational measures of self-confidence in L2 employed.
To what extent can self-ratings of one’s L2 proficiency be regarded as
solely reflecting self-confidence in L2? It seems highly probable that
when asked to self-rate one’s L2 proficiency, one would try to judge one’s
competency. This would imply that L2 proficiency rather than self-
confidence would play a definite if not a major role in the self-rating of
L2 proficiency. Hence the ratipgs hardly can be taken as a variable
explaining level of L2 proficiency. Operational problem aside, before
accepting self-confidence in L2 as an important motivational basis for
learning an L2, a first and foremost question seems to be how often
anxiety or confidence has been found to be related to L2 achievement.
Other studies in this area have included confidence or anxiety measures
and related them to L2 achievement. Gardner, Smythe, and ClCment
(1979) found that French classroom anxiety was related to one of the
three L2 achievement measures in their American sample but not in their
90 Language Learning Vol. 38, No. I
SUMMARY
NOTES
REFERENCES
Table A-1
Studies Relating Integrative Motive and L2 Achievement
Authors and Date Procedure Relevant results
Section 1
~~
Gardner & Lambert, 1959 75 grade 11 English Canadian high school Achievement was found to load highly (loading > .30) on both
pupils and 14 variables were used and the linguistic aptitude factor and on the motivation factor. These
results were factor analyzed into 4 two factors were found to be independent of each other.
orthogonal factors.
Gardner, 1960 83 grade 11 English Canadian students Achievement measures were found to load with linguistic and
learning French as an L2 and 30 variables integrative motive measures, Both orientation and desire to learn
were used. Results were factor analyzed into French were found to load with all 3 measures of linguistic
orthogonal factors. aptitude.
Lambert, Gardner, Barik, 192 students attending a 6-week French Correlation analysis revealed for the elementary group, both
& Tunstall, 1963 summer school as subjects. They were attitudes toward France and orientation to learn French
tested among other things on their attitudes correlated positively and significantly with achievement in
toward France, orientation to learning French. For the advanced group, orientation correlated with
French and French achievement. achievement in French whereas attitudes toward France were
found to be negatively and significantly correlated with French
achievement.
Gardner & Lambert, 1972: 96 American subjects and 41 variables were The factor matrix revealed one motivation factor and one
the Louisiana study used, 17 of which were French achievement attitude factor. The motivation factor received one high loading
measures. Factor analysis yielded 10 from a French achievement measure while the attitude factor did
orthogonal factors. not relate to either the motivation measures or any of the French
achievement measures.
the Maine study 145 American subjects and 41 variables A motivation and an attitude factor were yielded. The former
were used, 15 of which were French received high loading from one achievement measure while the
achievement measures. Factor analysis latter did not receive high loading from any of the 15
yielded orthogonal factors. achievement measures. Measures of integrativeness were not
found to load with measures of motivation.
Table A-I
Studies Relating Integrative Motive and L2 Achievement, continued
Authors and Date Procedure Relevant results
the Connecticut study 142 American subjects and 41 variables An attitude and a motivation factor were yielded. The attitude
were use, 15 of which were French factor received no appreciable loading from any of the 15
achievement measures. Factor analysis achievement measures while the motivation factor received one
yielded orthogonal factors. high achievement loading. Integrative orientation was negatively
related to 3 achievement measures. Measures of integrativeness
did not load substantially with measures of motivation.
the Louisiana study 80 high school French-American pupils and The integrative motive factor did not receive high loading from
41 variables 17 of which were French any of the 17 achievement measures. Attitudes toward
achievement measures. 12 orthogonal French-Americans loaded with one achievement measure in one
factors were yielded. factor and with 3 achievement measures and 1 aptitude measure
in another factor.
the Maine study 98 French-Americans and 41 variables were Desire to learn French and motivational intensity to learn French
used. 9 orthogonal factors were yielded. loaded with one achievement measure. The integrative motive
factor received high negative loadings from one MLAT measure
and one achievement measure.
the Philippine study 103 Philippine high school students learning 5 scales of the integrative motive loaded highly with two
English and a total of 47 variables were measures of English achievement. Instrumental orientation
used. Factor analysis yielded 7 orthogonal loaded with 2 English achievement measures.
factors.
Smythe, Stennett, & 2 samples of English Canadian subjects For the T-R sample, the orientation index loaded with 2 MLAT
Feenstra, 1972 learning English acted as subjects. One subtests and 5 measures of French achievement. A factor labeled
sample followed a traditional teaching the motivation factor received no high loading from French
program (T-R sample) the other followed achievement measures. For the A-L sample, the motivation
the audiolingual approach (A-L sample). 31 factor received high loadings from 3 French achievement
variables were factor analyzed into 5 measures.
orthogonal factors.
Gardner, Smythe, & 3 1 measures were administered to 62 Comparing the pre- and post-scores, it was found that subjects
Brunet, 1977 Canadians both before and after their had become significantly more ethnocentric, more highly
5-week intensive French course. motivated to learn French and had more opportunity to use
French. They became however, significantly less interested in
foreign languages and less integratively orientated. Their skills in
French were naturally found to improve throughout the course.
ClCment, Gardner, & A total of 304 grade 10 and 11 Montreal For the grade 10 sample, 3 achievement measures loaded with 3
Smythe, 1977 Francophone students were tested on a integrative measures. For the grade 11 sample, 2 English
number of integrative motive measures and achievement measures loaded with 4 integrative motive measures.
3 English achievement measures. Scores so The integrative motive factor which had a whole host of
obtained were factor analyzed separately for attitudinal and motivational measures loaded on did not receive
each sample yielding 4 orthogonal factors high loading from the English achievement measures.
for each sample.
Gardner, Smythe, & 2 samples, one American the other For the Canadian sample, no integrative motive measure was
ClCmenet, 1979 Canadian, of adult students following an found to load on the achievement factor. For the Americans, one
intensive French language program were achievement measure loaded negatively and highly on the
pre- and post-tested on a series of integrative motive factor. Comparing the pre- and post-test scores
attitudinal, motivational and French revealed that while some measures of the integrative motive
achievement measures. Factor analyzing received significantly higher scores, other measures of the
each sample separately yielded 3 orthogonal integrative motive received significantly lower scores. This was
factors for both samples. true for both samples.
CICment, Gardner, & 223 grade 11 Francophone students from Motivational intensity loaded with 2 measures of English
Smythe, 1980 Montreal were tested on a number of achievement. An integrative motive factor which received high
integrative motive measures and 4 English loadings from a number of integrative motive measures did not
achievement measures. Factor analysis receive substantial loadings from any of the English achievement
revealed 3 orthogonal factors. measures.
~~
Section 2
Anisfeld & Lambert, 1961 Scores obtained from measures of For 2 classes (A and B), the A-S score was negatively and
achievement in Hebrew were correlated with significantly correlated with achievement measure. Scores
scores on the Anti-Semitism (A-S) scale and obtained from the orientation index were found to be negatively
with that of the orientation index. Subjects and significantly correlated with achievement ratings in class C
were 81 Canadian Jewish children from 3 only indicating a positive relationship between instrumental
different Hebrew classes. orientation and achievement.
Table A-1
Studies Relating Integrative Motive and L2 Achievement, continued
Authors and Date Procedure Relevant results
Spolsky, 1969 4 groups of subjects totaling 315 students of The integrative motive as elicited by the Spolsky identity scales
diverse first language backgrounds learning was found to be positively and significantly correlated with
English in the U.S. were administered the English proficiently for all groups.
Spolsky identity scales.
Lukmani, 1972 60 Marathi speaking high school girls Ratings of the English speaking community did not correlate
learning English as a L2 in India were with cloze score. Cloze score correlated positively and
administered direct questions asking significantly with instrumental motivation but not with
subjects’ reasons for learning English as integrative motivation as measured by the direct questions.
well as the Spolsky identity scales. Subjects’
English proficiency was measured by a
cloze test.
Randhawa & Korpan, 1973 100 grades 7 and 8 Canadian pupils learning 30.6% of the total variance was found to be accounted for by all
French acted as subjects. Scores of an the 7 independent variables combined.
attitude toward learning French scale
specifically designed for use in this study
were factor analyzed into 4 orthogonal
factors. These factors together with 3
integrative measures were entered into a
regression equation using subjects’
achievement ratings as independent
variables.
Tucker, Hamayan, & Subjects were 74 Canadian grade 7 pupils. It was found that the factor which was defined by high loadings
Genesee, 1976 Scores obtained from a series of attitudinal from scales designed to measure subjects’ attitudes toward
and motivational measures adapted from French language and speakers came out as the significant
the AMTB were factor analyzed into 4 predictor in all 4 regression equations.
factors which were subsequently used as
predictor variables in 4 regression equations
using 4 different achievement measures as
dependent variables.
Lyczak, Fu, & Ho, 1976 210 1st year Chinese (Cantonese speaking) No significant correlation was obtained between the differential
students at the University of Hong Kong scores (ratings of the English guises minus ratings of Chinese
learning English acted as subjects. The guises) and achievement ratings in English.
matched-guise technique was used to elicit
attitudes toward learning English.
Oller, Baca, & Vigil, 1977 60 Mexican women learning English as a L2 8 factors were identified for the American ratings. Two entered
in a Job Corps were administered the the regression equation significantly. The 1st factor was
Spolsky identity scales and direct questions principally defined by both favorable as well as unfavorable
concerning reasons for learning English traits while the 2nd was primarily defined by favorable traits. 7
adapted from Gardner and Lambert (1972). orthogonal factors were distilled from the direct questions. One
Each of the 4 Spolsky scales and the direct labeled instrumental orientation entered the regression equation
questions were first factor analyzed and was found to be negatively related to L2 achievement.
individually and orthogonal factors
obtained entered into 5 different equations
using subjects’ cloze scores as dependent
variables.
Oller, Hudson, & Liu, 44 native speakers of Chinese university Subjects had high English proficiency if they rated Americans
1977 students in America were administered the high on the 2 factors labeled Americans as successful, Americans
Spolsky identity scales as well direct as helpful and low on the 2 factors labeled Americans as
questions concerning subjects’ reasons for sensitive and Americans as happy. The 1st two and the 4th
coming to the U.S., their attitudes toward factors contained all favorable traits while the 3rd had both
the U.S., and their reasons for learning favorable and unfavorable traits loaded on. Results obtained
English. Their English proficiency was from the direct questions indicated those who had low doze
measured by a doze test. Exactly the same scores were inclined to stay permanently in the U.S., thought
procedures employed in Oller, Baca, & their job opportunities were better in the U.S., and had long
Vigil’s 1977 study were used to analyze planned to come to the U.S.
results.
Chihara 62 Oller, 1978 The methodology employed was identical to The higher subjects rated English people as confident and
the previous study. Subjects were 123 broadminded, the lower their doze scores. Both confident and
Japanese adults learning English in Japan. broadminded were favorable traits. The higher they rated English
people as modest and shy, the lower their cloze scores. The
former was a favorable trait, the latter was an unfavorable trait.
Table A-1
Studies Relating Integrative Motive and L2 Achievement, continued
Authors and Date Procedure Relevant results
Pierson, Fu, & Lee, 1980 466 secondary school pupils, all native Results obtained from the direct questions revealed that the more
speakers of Chinese learning English in freedom subjects felt they should have in choosing the language
Hong Kong, were administered 23 attitude for instruction, the stronger their desire to learn English, the less
statements. Ratings obtained were factor approval they felt they should receive from others in using
analyzed into 11 factors, 6 of which entered English, the less discomfort they felt about other Chinese using
a regression equation significantly using English and the less they felt English was the mark of an
their cloze scores as the predicted variable. educated person, the higher their cloze scores. Results from the
The Spolsky identity scales were also Spolsky identity scales indicated that the more subjects felt
administered and factor analyzed separately. Western people could be described as gentle, graceful, and
Factors so obtained were entered into 4 trustworthy, the lower their cloze scores. The more subjects felt
different regression equations using Western people could be described as logically minded and wise,
subjects’ cloze score as the dependent hard-working and self-confident, the higher subjects’ cloze
variable. scores. All 5 traits were favorable characteristics however.
Muchnick & Wolfe, 1982 21 classes of high school students of 4 orthogonal factors were yielded none of which was found to
Spanish were administered all the scales of correlate significantly with grade in Spanish.
the AMTB adopted for use in this study.
Wong, 1982 50 native Chinese adolescents learning The scales were found to yield high test-retest reliability
English acted as subjects. The AMTB was coefficients (average .91). No significant correlation between
adapted to measure subjects’ motivation to scores on the AMTB and that of an English achievement test was
learn English, orientation to learn English, obtained.
desire to learn English and attitudes toward
English.
Au, 1984 Two samples (n = 59, n = 77) of ethnic Desire to learn English was found to load negatively with English
Chinese (Cantonese) speakers of English proficiency measures both in the 1st test and retest of one of the
were tested on a number of integrative samples (n= 77). High confidence in using English was found to
motive and other measures. One of the relate to high English proficiency in all samples. Using the
samples (n = 77) was retested 6 months later regression technique on the longitudinal results revealed however
in order to test the causality hypothesis. that no causal statements could be made with regard to
Two English achievement measures were integrative motive variables and English achievement.
used. Factor analysis and regression analysis
techniques were used to analyze results.