Investigating Language ClassAnxiety

Using the FocusedEssayTechnique

PETER D. MACINTYRE and R. C. GARDNER
Departmentof Prychokglt Depa ment of Pslchologl'
Unirers;tt oJ Westem Ontaio Universi| oJ Wettzm Ontario
London, Ont, CanadaN6A 5C2 Lo itn, Ont., CanadnN6A 5C2

FORMANY [Link] Beyond the effect of anxiety on performance

be more anxiety-provoking than any other in the secondlanguage, it appearsthat anxious
course that they take (1; 13; 20).t Anecdotal students experiencebasic problems in vocabu-
reports and an emerging body of empirical lary acquisition and retrieval. We have in-
research suggestthat language anxiety can have [Link] a controlledlaboratory setting.
a pervasive impact at all stages of language the relation of French vocabulary learning and
learning and production (14; 19). production to several forms of anxiety (20). A
In the language classroom, increased levels factor analysis [Link] nine dillerent anxi-
of arxiety can have a variety of negativeeffects. ety scalescould be represenredby two main
Anxious students are less likely to volunteer anxiety dimensions'. Gened Anxiety and. Com-
answersand to participate in oral classroomac- municatiaeAnrieQ. The Gmtal Anxietl factor was
tivities (5). They also tend to avoid difficult primarily defined by anxiety unrelated to lan-
linguistic structures that the more relaxed guage or speaking and scoreson this factor were
studentswould be willing to attempt (16). Not not related to the learning or production of
surprisingly, therefore, Ianguage anxiety has French vocabulary . Communicatioe Anxietl was
been negativelycorrelatedwith languagecourse defined by anxiety in interpersonal settings,
grades ( 12) and teachefs ratings of achievement especiallyFrench speaking situations, and was
(33). significantly related to both the learning and
While anxiety is correlated with these more recall of vocabulary items. More detailed
global measures, its negative influence on the analysisdetermined that the presenceofthe two
specificaspectsof the languagelearning pro- measuresofFrench-related anxiety on this fac-
cessmust also be [Link] has been tor likely accountedfor its associationwith per-
found to influence listening comprehension formance in French. Therefore, students who
negatively (10). Anxious students will some- are a-nxiousin French speakingsituationsap-
times report that language class moves too pear to be disadvantaged from the outset
quickly, that they feel left behind, and that they becausebasic vocabulary learning and produc-
require more time for their studies- a fairly tion is impaired by the apprehension they
c o m m o n r e s p o n s et o a n x i e t y i n a n y i n s t r u c - experience.
tional setting (32). Word production of anxious ln a subsequentstudy. rwenty-oneanxiety
students also tends to be smaller than that of scaleswere factor analyzed to determine their
their more rela-redcounterparts (10) and their underlying dimensionality (18). A solution with
speechtends to be lesscomplex and interpretive three factors was retained. Four measures of
(31). Such difficulties can lead to the impres- language anxiety were included in the study
sion that anxious studentsare not capablecom- and these scalesformed a factor of their own
municators in the secondlanguage. The nega- called LanguageAnxr'el_/.This factor was distinct
tive implications also extend to tests and ex- from both a Social Eoaluation An iet) lactor, as
aminations where the performance of anxious defined by ego-threat anxieties such as com-
students may be limited (35). munication apprehension, a\d a StateAn iet!
factor, as defined by anxiety at particular
moments during the study. This study also in-
TheModunLanguseJou/nal,75, lii (1991) cluded measuresof inirial cognitiveprocessing
0026-7902/9 | /0003/296 $1.50/0
@1,991 (digit span) and retrieval from long-term
TheModemLans@ge Journal
memory (vocabulary production) in both the
PeterMacin\/re U R. C. Gardner 297
native language (English) and the second can be aware that only in language class do they
Ianguage(French). When aggregatedvariables experiencedi{ficulty. This leads to the differen-
were createdto representeach ofthe three aru<- tiation of language anxiety from other forms
iety factors, only LanguageAnxie$ cortelated of anxiety and to the isolation of its effects in
significantly with initial processing a:rd retrieva.l the second language context.
from memory, and then only in the second Thus far, this model has been tested among
language. more experienced language learners. A recent
These two studies also demonstrate the fun- review ofthe literature indicates that almost all
damental importance of the type of anxiety studies of language anxiery examine it as a
under consideration. In both studies, several stablepersonality trait, among experiencedlan-
different types of anxiety were examined, in- guage learners. Its origins and development
cluding test anxiety, audience anxiety, and have receivedmuch lessattention (19). There-
math classanxiety among others. Only in cases Ibre, researchis required that directs attentron
where the anxiety was specifically related to a to the nature of language anxiety in its earlier
second language context was there clearly a stages among less experienced students. In
relationship between anxiety and second lan- terms of our model, the focus would be on the
guage performance. This distinction between stage at which anxiety is not well differentiated.
languageanxiety and other forms ofanxiety has Such researchwould be valuable in that it could
been made for some time and is now beinE suggestways in which the development ofthis
recognizedas a key issuein the understanding anxiety may be curtailed or its effects amelio-
of the role of anxiety in language learning (B; rated.
13; 20). One may approach the problem ofreducing
We have suggesteda model that attempts to the negative effectsofanxiety as an attempt to
describethe processby which language anxiety change students' self-related cognitions. In a
becomes differentiated from other forms of similar vein, studies ofattitude change suggest
anxiety (20). We propose that initially, anxiety that by requiring an individual to think about a
is an undifferentiated, negative affective specific aspectof a given attitude, that attitude
responseto some experiencein language class. can be [Link],under certain conditions (7). For
With repeated occurrences, anxiety becomes example, one study found that it was possible
reliably associatedwith the language classand to bias self-evaluations along a dimension of
differentiated from other contexts. Therefore, "environmentalism" by slight changes in
the
we can obtain near zero correlations between, wording of attitude statements (2). These
for example, anxiety in language class and changesmade it easier or more difficult to en-
anxiety in mathematics class or observe dif- dorse pro-environmentalist statementsthereby
Grent patternsof correlationsbetweenlanguage directing subjects' perceptions of themselves
a c h i e r e m e nat n d r a r i o u s t y p e so f a n x i e r y . toward or away from environmentalist atti-
Differences between more and lcss anxrous tudes. The subjects saw themselvesendorsing
students can be traced back to the manner ln either more or less "environmentalist"
which anxiety is presumedto exert its negative statements and therefore saw themselves as
effects. The experience of anxiety has been m o r e o r l e s so f a n e n r i r o n m e n t a l i s r .
discussedin terms ofcognitive interferencepro- Our model oflanguage anxiety suggeststhat
duced by self-relatedcognitions (27). Anxious this may be possible.e"peciallyamong inex-
individuals think about their own reactron to perienced language students. If anxious
a task in addition to the demands of the task s t u d e n t sr o u l d f o ,u s o n p g r 1 1 1 u . . * p e r i e n ci e
ns
itself (6). The content of their thoughts is t h e s e c o n dl a n g u a g er. a r h e rt h a n o n n e g a t i v e
negative and centred on self-degradation (e.g. ones, the d€bilitating effects of language anx-
" l m j u s t n o g o o d a t t h i s " 1 .L e s " a n x i o u sp e r - iety could be reduced.
sons do not have such exaggeratedself-aware- The present study attempted to manipulate
ness and can therefore concentrate more fully the anxiety level ofbeginning languagelearners
on the task itself(32). In the caseoflanguage by asking them to think about and report either
anxiety, studentsin a language classmay have p o " i t i v eo r n e g a t i v ee v e n t sf r o m r h e i ro w n e x -
negative thoughts that do not intrude on their perience. These half-page "focused essays"
work in other [Link] studentsengage forced the studentsto concentrateon their own
in rumination over poor performance that reactions to events making a given type of ex-
limits tbeir ability ro proces" information in perience higlrly salient ,o ,rzr,?.By writing about
secondlanguage situations. Thus, the students one type of experience, it was anticipated that
294 The Modern LanguageJoumal 75 (1991)
students' perceptions of themselves would and confident or v€ry nervous and apprehen-
becomebiasedin the direction suggestedby the sive. Each subjectwas askedto write a half-page
essaytopic. essay on one of these experiences. Approxi-
Analysis of the brief essaysthemselves can mately half of the students were randomly
provide information on the development of assignedto write an anxiety essay(n = 16) and
anxiety and those specific experiences that half to write a relaxed essay (n = 15).
students find lrrost anxiety-provoking or most Can Do. Twenty-eight items from Clark (3)
conhdence-building. In addition, the effect of were chosen to reflect a range of French tasks
essay writing and anxiety levels on language requiring various ability levels. The abilities of
performance tasks, in both the native and sec- all of the students were unknown at the outset
ond languages, also will be examined. of the study, therefore a range of items repre-
senringvery simple tasks(such as counting to
ten) to moderately difhcult items (such as
METHOD
understanding a conversation between two
[Link] intact classesfrom the continu- native speakers) was selected. Respondents
ing education divisions ofa community college rated their ability to do each of the tasks on a
a n d a l a r g eu n i v e r s i t la g r e e dr o p a r t i c i p a t ei n seven point scalewith the anchors "very difficult
the study. These introductory, non-credit for me" and "very easyfor me.'Four sub-scales
classes(Level I) were all designed to introduce comprise the overall measure: speaking (eight
conversational French skills to adults. The items), comprehension (eight items), writing
classeswere relatively small and a final sample (six items), and reading (six items).
of thirty-nine adult second language students Production?Zrls. Four measuresoflanguage
were tested during an hour of regular class performance were taken, two in the native
time. language and two in the second language:
Eight studentshad to be eliminated liom the English digit span, English vocabulary produc-
sample becausethey did not comply with the tion, French digit span, and French vocabulary
instructions. The remaining students had an production. The Digit Spanincluded strings of
average of only 3.1 years of previous French numbers between four and nine digits in length
education at some time in elementary or sec- which were administered first in English and
ondary school and reported an averageo1 14.1 then in French. A measure of vocabulary was
years sincethey had studied French. Therefore also taken as the response to a Thing Categorl
the sample may be described as being com- ?sl which required the naming of elements of
prised of inexperienced language learners. By six categories,such as words beginning with the
" t " o r i t e m s t h a t b e l o n gi n a s u i t c a s e .
[Link] subjectsin one ofour previous l e t t e r
studies had an average of B.1 years of French- Three categorieswere answeredin English and
as-a'second-languageeducation and all had three were answered in French.
studied French within the past two years. [Link] Anxometers (anxi€ty ther-
Materiak. Six anxiety scaleswere chosen in mometers)were administered in order to assess
order to span the range of anxieties that have the amount of anxiety created by each of the
been included in previous investigations. The four production tasks. These measures are
scales were: French UseAnxietl and FrenchClass single-item, visual analog scaleswith a range
Anxietl (21), Pnsonnl ReportoJCommunication Ap- ofvalues from one to ten printed along the side
prehension(23), Fear oJNegatioeEmluation (34), of a thermometer frgure (see 18 for an ex-
Tiait Awitty (15), and StateAnxiet! (29). All of ample). Subjects are asked to draw a line on
the anxiety scalescontained an equal number the Anxometer to indicate their level of anx-
of items keyed positively and negatively for iety during each of the four production tasks.
anxiety and were answered on a seven point Procedure. As subjects arrived for class, they
Likert scale,with the exception ofthe StateAM- were each given one of two forms of the ques-
zat_2questionnaire which employed its usual tionnaire. The forms were [Link]
four-point scale. the type of essayrequested. The booklet was
In addition, subjectswere askedto compl€te arranged such that all ofthe anxiety scalesap-
the "Essay,""Can Do," "Production Tasks,' and peared in the first section with their items ran-
" A n x o m e r e r sp' o r r i o n so f t h e q u e s t i o n n a i r e . domly mixed, followed by writing of the Essay,
-[Link] were asked to recall an ex- the StateAnxiet! Scale, the Can Do scale, pages
perience that required the use of their French on which to respondto the ProductionTixks, atad
skills during which they felt either very relaxed finally the lour Ar*ometers.
Pttor Ma.c;nttre €4 R. C. Gardna 299
Subjects were given as much time as re- sevenpercent). Recall of an anxiety-provoking
quired to complete the self-report sections of experience is highly likely to involve diffrculty
the questionnaire. When all were finished, the with speaking skills.
production tasks were administered by tape SubjectiaeRatings oJ [Link] order to
recorder. Subjects then were asked to assess assessthe effects of essaywriting on the sub-
their own anxiety reactions to each ofthe tasks .jective ratings of proficiency, t-tests were con-
using the Anxometers. ducted on the Can Do scores for each of the
four abilities; speaking, understanding, read-
RESULTS ing, and writing. The only significant difference
was obtained for the self-ratedspeakingability,
The results of this study can be considered
with those writing a relaxed essayconsidering
in two broad sections. The first will compare
themselves as being more proficient than the
the two essaygroups. Content analysis of the
anxious essay group (see Table I).
lhemes of the essaysrevealed rhat anxious
Objectiae [Link] differenceswere found
essayswere almost exclusively descriptive of
between those writing about a rela-xedexperi-
speaking events while conhdent essays were
enceand thosedescribingar an-riousexperience
based on both speaking and understanding
on the Digit Span or Thing Category Test
skills. Also, the essaysappearedto influence the
scores in either English or in French (all t's
self-ratings of proficiency while the objective
( 1.60, seeTable I). Writing the essaydid not
measures did not differ between the groups.
appear to differentially affect the objective
The secondset ofresults will examine the in-
measures that followed it.
fluence of previous anxiety on both the
Antiery Reactions to 1Zr*s. When the two
language performance measuresand the reac-
groups are compared for differences in State
tions to the tasks. In both cases, differences
Anxiety. the differente ber,,reenrhe groups is
were found based on levels of anxiety, type of
not significant (seeTable I). Thus, it would ap-
task, and the language of the task.
p e a r t h a t s i m p l yw r i t i n g r h . r s \ a y r ^ a sn o t . i n
itself, anxiety provoking or relaxing for either
COMPARISONS
BETWEENESSAYGROUPS
group. Moreover, the differencesin the scores
Essay [Link] of the students were on each ofthe four Anxometers corresponding
unable to write an essayon the requested type to those measuresare also not significantly dif-
of experience. Three could not recall an anx- ferent (see Table I). This would indicate that
ious experience, five could not recall a confi-
dence-building one. All of these data were, TABLE I
therefore, excluded from the analyses,leaving Effects of Essay Condition on a1l Dependent Variables
the final sample of thiny-one students. No
significant differences were found on any Relaxed Anxious
measure between those subjects who were in- Mean Mean t
cluded in the analysesand those who were ex- Se{- Ratiner of Ptofic;ner
cluded (all t's ( 1.4). Speaking 3 7. 3 29.8 2.85.
The essayswere read independently by two Understanding 28.9 26.3 1.44
judges who attempted to classify the dominant Reading 23.r 19.1 r.66
theme ofthe essayas requiring speaking, com- Writing 21.9 18.6 1.45
prehension, reading, or writing ablity (100%
agreement). The essays,however, actually fit Digit Span

only two of the four potential classifications: English 54.9 55.2 0.07
Frcnch 44.9 43.3 0.34
spealingand comprehension. A chi squarete.t
Categories
ofindependence indicated that the type of skill 32.6 1.60
English 2?.t
reporled in the essaywas dependenton the alx- French 16.3 16.0 0.13
iety group to which the subject had been as-
s i g n e d( 1 ' ( 1 ) - 3 . 9 7 , p ( . 0 5 ) . T h e r e l a x e d 34.8 35.0 0.07
essaygroup wrote about having confidence in
their comprehension skills about as olten as Digit Span- English 4.6 4.5 0.09
they wrote about speaking skills (forty-seven Digit Span - French 6.6 6.3 0.34
and fifty-three percent respectively). On the Categories- English 3.9 3.6 0.33
Categories-French 5.8 5.0 1.00
other hand, the anxious essay group wrote
almost exclusivelyabout speakingskills (eighty- Note: *p<.05.
300 Joumal 75 (1991)
Ttu Motum Langunge

the anxiety aroused by the performance on each ing analysis. The resultsshow that a significant
of these tasks also was not affected by the type main effect was obtained only for Language
of essaythat was written. [F(1,29)= 106.89,p<.01]. Figure I showsthat
In summary, therefore, examination of the the mean categoriesscoresare higher in English
content ofthe essaysrevealedthat, when called than in French. The differences obtained for
upon to rePorl anxiety-arousingexperiences. the French Anxiety factor are in the expected
direction, but are not significant (F(1,29):
s t u d e n t sf o c u sa l m o s te x c l u s l v e lol n s i t u a l i o n s
involving speal<[Link], on the other 1.37, n.s.). The interactionis also not signifi-
hand, seemsto come equally from the ability c a n t ( F ( 1 , 2 9 )= . 3 2 , n . s . ) .
to successfully comprehend a message or to The above analyses conducted with the
deliver one. The type ofessay written appears French Anxiety factor were also conducted with
to affect the self-ratingsof speakingproficiency a different anxiety factor in order to examine
but not th€ more objective measures of profi- the importance of the type of anxiety in terms
ciency. Further. the essaydid not seemto in- of its effects on the performance measures. A
fluence the level of anxiety experienced by the median split on the aggregateof the State and
subjectswhen being tested. Becausethe essay Trait anxiety scaleswas created as an index of
manipulation did not inlluence the scoresofthe General Anxiety. The results ofthese analyses
respondents on the production measures, the are very similar to the results observed for the
analyses to follow can be performed without French Anxiety analyses (see Figure II).
considering the respondent's essay group. For the digit span tasks, main effects were
observed for General Anxiety [F(1,29) = 4.90,
p < . 0 5 1 ,a n d L a n g u a g e[ F ( 1 , 2 9 )= 2 5 . 8 2 ,p <
ANXIETY AND PRODUCTION .01], while the interaction was not significant
In order to examine the effects of previous l F ( 1 . 2 9 )= . 3 2 , n s . l F o r t h e c a t e g o r i essc o r e s '
anxiety and language of administration <lnthe the only significant effect was for Language
four performance measures, two separate [ F ( 1 , 2 9 ) =1 0 6 . 1 ,p < . 0 1 ] . A c o m p a r i s o no f
analysesofvariance were conducted. The first Figures I and II will illustrate the striking simi-
analysis examined the effects of French Anx- larities between these analyses. This finding
iety and Language on the Digit Span scores seemsto indicate that the source ofthe anxiety
while the second assessedthe effects on the does not matter in this sample. That is, con-
Thing Category Test scores. To classily stu- trary to the conclusions of previous studies, it
dents as high or low in French Anxiety, scores would appear that situation-specifrc French
on the French Class and French Use anxiety Anxiety affects performance on French tasks in
scaleswere added together and the total was the same manner as cross-situational General
split at the median. Anxrety.
The scoreson both the English and French The [Link] anxieLyreac-
Digit Span tasks were subjected to a 2x2 tions to the various production tasks. A
Analysis oI Variance with the factors being 2x2x2 Repeated Measures Analysis of
French Anxiety (high or low) and Language Variance was perforrned on the Anxometer
(English and French). The results showed sig- scoreswith the factors French Anxiety (high or
nificant main effects for both French Anxiety Iow), Language (English and French), and
[f(1,29) = 5 . 4 2 , p < . 0 5 ] a n d L a n g u a g e Task (digits and categories). Signifrcant main
[F(1,29) : 29.33, p ( .01], however,the inter- effects were obtained for French Anxiety
action between the two was not significant l F ( 1 , 2 9 ) : 5 . 5 4 ,p < . 0 1 1 L , a n g u a g e[ F ( 1 , 2 9 ) :
= .02, n.s l Figure I shows that the 5 . 4 8 , p < . 0 1 1 , a n d T a sk [F(1,29):22.90,
tF(1,29)
Low French Anxiety group had higher mean p< .011,while none of the interactionswere
digit span scoresthan those with High French significant (all F's <.60). As shown in Figure
Anxiety. Figure I also indicates that the Digit I I I . h i g h l y a n x i o u ss l u d e n t sf e l t m o r e a n x i e t y
Span scores were higher in English than in during all ofthe tasks than did the low anxiety
French. group. Also, the tasks in French aroused sig-
The scoreson both the English and French nificantly more anxiety than they did in Eng-
Thing Category Tests were also subjected to a lish. Finally, it can be seen that the digit span
2 x 2 Analysis of Variance with the French test was more anxiety provoking than was the
Anxiety and Language factors as in the preced- catesorles test.
Petct Macintyre 6 R. C. Gardner 301

FIGURE I
LanguageProductionScoresin Native and SecondLanguagesfor the Two FrenchAnxiety Groups

score

French E n gl i s h French English

DIGITS CATEGORIES
I Low French Anxietv % H i g hF r e n c hA n x i e t y

FIGURE II
LansxageProductionScoresin Native and SecondLanguagesfor the Two GeneralAnxiety Groups.

score

French E n gl i s h French E n gl i s h
DIGITS CATEGORIES
I Low Anxiety % XrgnAnxiety
302 The Moden LanguageJoumal 75 (1991)

FIGURE III
Anxometer Scores for Each Type of Task, in Each Lmguage, for High and Low French Anxiety croups.

Anxometerscore

French E n gl i s h French English

DIGITS CATEGORIES
I tow anxiety % nign anxiety

A similar analysis was also conducted with iety in the language classroom are at a disad-
General Anxiety in place of French Anxiety. vantage when compared to their more relaxed
The significant main effects were again ob- colleagues.
tained for Language and Task because these Somewharcontradictory r,rith previous re-
factorswere the sarneas the preceding analysis. search is the relationship observed between
In this case,however, neither the General An-.<- General Anxiety and the French production
iety effect nor any of rhe inleracrionswere tasks. Several studies have indicated that
significant. Language Anxiety can be distinguished from
General Anxiety, both theoretically and em-
pirically (13; 19). In two earlier studies(18; 20),
DISCUSSION we found that only those scalesdirectly related
S o m er e s u l r so f t h i s s t u d y a r e q u i r e c o n s i s - to anxiety in second language contexts were
tent with those obtained in the literature on related to performance on second language
Language Anxiety. Previous studies have tasks. In the present study, both French Anx-
shown that those with higher levels of French iety and General Anxiety demonstrated the
Anxiety tend to perform more poorly than same relation to scores for the French tasks.
those who are Iess anxious (19). Also, those Also in contradiction to previous results (18),
with higher levelsof French Anxiety tend to ex- French Anxiety was associatedwith impaired
periencemor€ anxiety during the activities that performance on the English tasks. Thus it
involve the use of French (18). Both of these would appear, at first glance, that the discrimi-
resultshave been replicaredin rhis investiga- nant validity ofthe different anxiety constr-ucts
tion. LanguageAnxiety consistently,negatively is questionable,
affects language learning and production. Our model (20) may help to explain these
Therefore, studentswho often experlenceanx- apparent [Link] beginning language
PeterMacintlre I R. C. Gardner 303
learners would be expected to €xpenence un- suggestthat if teachers can promote the more
differentiated anxiety becausetheir experiences positive speaking experiences rather than the
in language class have not had sufficient time anxiety-provoking ones, their studentswill feel
to become reliably discriminated from other better and may learn nore elficiently as well.
types of anxiety experiences. This is reflected This is certainly a difficult decision to make a
in the similar effects of General Anxiety and priori, although some suggestionsIbr promot-
French Anxiety on the performance measures ing confidence-building experienceshave been
in both languages. Unfortunately, the nature made(1; 17).
ofthe studenra s n d r h e c o u r s ed o c sn o r p e r r n i t For this reason it is encouraging that the
a follow-up testing. However, based on the focusedessaywas successfulin altering the self
results ofthe studies previously cited, it would perceptions of the language students. Those
be expectedthat, as the studentsprogress,only who wrote a relaxed essayindicated more con-
anxiety specifically related to secondlanguage f i d e n re w h i l e r h o s ew r i t i n g a n a n x i o u se s s a y
learning would be associatedwith second lan- showed less confidence with their language
[Link] processmay be ma:ri- skills on the Can Do scale. This is a rather
fest in the reactions to the tasks, as measured subtle manipulation, but it would be the kind
bv the Anxometer [Link] a differentia- of reflection students could perform spon-
tion of the effects of French Anxiety and taneously. Conceivably, studentstaught to em-
General Anxiety was observed. phasize their own successfulexperiencesin the
Although the anxiety rnay not yet be fully secondlanguagewould come to perceivethem-
differentiated, anxiety in French class is a selvesas more proficient language learners, in-
meaningful experience, as demonstrated by creasing their self-confidence(4). This would
Anxometer scoresand the content ofthe essays. take the focus away from the secondlanguage
The content analysis reveals that speaking is tasks and direct it toward reactions to those
the most anxiety-provoking of secondlanguage tasks, possibly making the language teacher's
activities. The frustration experienced when a job a little easier.
secondlanguagesrudenris unableto communi- The observed differences in the self-confi-
cate a messagecan lead to apprehensionabout denceofthe two essaygroups may be prompted
future attempts to communicate (13; 221,26). by changes in the self-related cognition of the
This would explain why anxious students tend students. Such a hypothesis would explain the
to avoid classroom participation (5). Con- differencesin the self-perceptionof proficiency
versely, when describing a confidence-building observed between the two essaygroups. That
experience, students concentrated on both effect cannot be explained by postulating that
understanding and speaking skills. Thus, as students in the relaxed group actually relaxed
suggestedby other studies,it would appear that or that the anxious essayincreased the levels
second language communication may be the of anxiety since there were no differences in
key determinantoI the affectiveexperiencein staLeanxiery berween the groups.
language class (20). Severalstudiesindicatethat self-perceptions
This represents a considerable challenge to can have a profound effect on behaviour (7).
conscientiouslanguage [Link] this study, While changing a person'sself-perceptionis not
the anxiety provoking experiencesdescribedby easy, the benefits of improving the self-image
the eighty-seven perc€nt of the students in- of language students seem worthwhile. Based
volved speakingin the [Link] the on the results observed here, it seemsreason-
other hand, approximately half ofthe reported able to assumethat, given time and a focus on
confidence-building experiences involved the positive experiences,we may encourage more
use of speaking skills. Clearly, then, the data self-confident and capable language learners.

sistance in cooidinating th€ testing potions of this study,

to the students demselves for their wilingness to panicipatc,
to A. Maclntyre for her assistancein the preparation of
this manuscript, and to the Social Science and Research
'The authors would like to express sincere gratitude to Council of Canada for financial support of this research
P. Watts at the Univenity of Westem Ontario andJ. Jensen (Felowship No. 4s3'90 0177 and Grant No. 410-90-0195).
at Fanshawe College in London, Ontario, for their as-
304 The Modzm Langnge Joumal 75 (1991)
17. Lucas, J. 'Comnunication Apprehension in the ESL
BIBLIOGRAPHY Classroom: Getting Our Students to Talk." Folrrgt
Lnguagc Anmts 17 (1984): 593-98.
"Language
18. Maclntyre, Peter D. & R. C. Gardner.
1. Campbel, Christine M. &Jos€ Oniz.'Helping Students Anxi€ty: Its Relation to Other Anxieties and to Pro-
Overcome Foreign Language Anxiety: A Foreign cessing in Native and SecondLangages." Languagc
Language Anxiety Workshop." Languag. AMtctJ: Irarn mg (forthcoming).
Fron Thcot), and Researchto Cksnoam hadicz. Ed 19. - & R. C Gadner. "Methods and Results in the
Elaine K. HoNitz & DollyJ. Youns. Englewood Study of Foreign Language Anxi€ty: A R€view of
C l i f f s , N J : P r e n t i c eH a l l , 1 9 9 1 r 1 5 3 - 6 8 . tl\e Llteraturre." language Luming 4t (1991): a5-lt7 .
'Affective-Cognitive .Anxi€ty and Second
2. Chaiken, Shelly & M. W Baldwin. 20. _ & R. C. Gardner.
Consistency and the Effect ofsalient Behavioral ln- Language Learning: To ard a Theoretical
formation on the Self'Perception of Attitude.'Jroznal Cclarirc^rlon.' [Link] 32 (1989): 251 75.
oJ Pusonatitl and Sociat Psychoks 4r (rs9l): | 12. 21 -&R C. Gardner. .The Measurement of AB-
"Languase." A Sunq of ctobat
3. Clafk, John L. D. iety and Applications to Second Language Learn-
Undd'tandine: F;Ml R porl . Ed. T. S. Banows. New ing: An Annotated Bibliogtuphy." Rsearch Bulbtia
Rochelle, NY: ChaneeM^gazine, 1981: 25-35. 672. London, Ontaio: Univ. of Western Onta o,
"Second Language Proficiency and
4. Cl6ment, Richard. 1988.
Acculturation: An lnv€stigation of the Effects of 22. McCroskey, James C. An Introdut;o to Rhetoia Con-
Language Status md Individual Chamcteristics." nunbation. Englewood Cliffs, NJ: Prentice Hall,
]oufml oJ Lansvase and Sociat Pychohs 5 (1987). 1982.
271-90. 23. -. "Validity of the PRCA as an Index of Orat
"An Analysis of Discomfort,
5. Ely, Chritopher M. Communication Apprehension." Comnuniation
Riskta-king, Sociability, and Motivation in the L2 M o n o e r a q h4s5 ( 1 9 7 8 ) : 1 9 2 - 2 0 3 .
Classroom." LanguaseLeam;ne 36 (1986): 1 25. "Iest Anxiety, Worry, ard Cognitive
24. Saason, Irwin G.
"Anxiety, Learning and Memory: A
6. Eysenck, Michael. Interfercnce." Sef Rekted Cogn;tin;n AM;'t) and
Reconc€ptualization.'Jronnal [Link];n Pnsonalitl Matitation. E'd. R. Schwarzer. Hinsdale, NJ:
13 (1979):363 8s. Erlbaum, 1986r 19-34.
"Self-Perception Theory: A Curlent
7. Fazio, Russel. 25. - Ian Anxiett: Thcar, Rcsrarch, and Inpkat;on'.
Perspective." Social Influen ): The Ontaio S|nposium [Link], NJ: Erlbaum, 1980.
(Vol. 5). Ed. Mark P. Zanna, James M. Olson, & "Social Anxiety and
26. Schlenker, B. R. & M. R. Leary.
C. P. Hermm. Hinsdale, NJ: Erlbaum, 19871 Commnnication about the Self ." Jounal oJLaEMgc
129-50. and Sotiat Pslehotos 4 (198s): 171-92.
8. Gardner, R. C. [Link] Pguhobg and Suond LanguageLam' 27. Schwarzer, Ra\f. Sef Rehted Casnition;n Anx;ett aad
ing. Thz RolcoJAttitu&s andMot;Mt;an. London:Ar' Motintion. r{illsda],, NJ: Erlbaum, 1986.
nold, 1985. "Ihe Effect of Affect on Foreign
28. Scovel, Thomas.
(Learnins
9 - Another Languager A True Social Language Learning: A Review of the Anxiely
Psychological Experiment." Journal of Languageatd Resear,h. Zargkqr Leamins 28 (1o78): 129-42.
Suial Pslchatnsy2 (1983): 219-39. 29. Spielberger, Charles D. Manual;for theState-TruitAM-
10. -, R. N. Lalonde, R. Moorcrcft & L T. Even. i'0 In"entoD,(Fam t. Palo Alto, Califomia: Con-
"Second Language Attrition: The Role of Motiva-
sulting Psychologists, 1983.
tior arduse.".Joulnol of Languge and SocialPqycholns 'r0 - "Complex Learning and Academic Achieve-
6 (1987): 29-47 . ment." [Link] and Behaur Ed. C. D. Spielberger.
"An Instrumental
11. - & Peter D. Maclnryre- New York: Academic, 1966: 361-98.
Motivation in Language Study: Who Says It Isn't "The Effect of
31. Steinberg, F. S., & Elaine K. Horwitz.
Effective?" Std,rr in [Link] 13 Induced Ani€ty on the Denotative and Interpretive
(1991).57-72. Content ofsecond Langu age Speech.' TESOL Quat
"Preliminary Evidence for the
12. Hor,r'{itz, Elaine K. terQ 20 (1986): 13r-36.
Reliability md Validity of a Foreign LaDsxage Anx- 'Auiety and Cognitive Processing
32. Tobias, Sigmund.
iety Scalc." TESOZ Qua eb 20 (1986): 559-62. of Instruction." Szf Ralatzd Cogn;t;on in [Link] and
"Foreisn
13 -, M. B. Horwitz & Joanne Cope. Mot;Mt;on. Ed. R. Schwarzer. Hillsdate, NJ:
Langtrage Classroom Anxiety." Modeln Lansuase Erlbaum, 1986:35-54.
Jounat ?0 (1986): r25 32. 33. Trylong, Vicki L. "Aptitude, Attitudes, and Anxiety:
14. - & DoIlyJ. Yo|ung. LaneuaeeAM;aU: Flan Thror] A Study of their Relationships to Achiev€ment in
and Researchto Cla-ssroonPrz.,;,. Englewood ClilTs, the Forcign Languasc Classroom.'Diss., Purdue
NJ: Prentice Hall, 1991. Univ, 1987.
"Interpretels Guide to theJackson "Measurement of Socia]'
15. Ja&son, Douglas N. 34. Watson, D. and R. Friend
Personality lnventory.' Adoancxin Prtcholog;cal As Evaluative Auiety."Jrdumal o;fConsultingand Clin;eal
rr$r?a?[Link]. P. McReynolds. San Fmncisco:Jossey- Pychologr 33 (1969)r aa8-51.
Bas, 1978:56-102. "The Relationship between Anxiety
35. Young, Dolly J.
'Avoidance Rehaviour in
16. Kleinmann, Howard H. and loreisn Languasc Oral Proflciency Ratings."
Adult Second Language Acquisition." Language ForeignLanguageAnnals 19 (1986): 439-45.
L m m i n g 2 7 ( 1 9 7 7 ) :9 3 - 1 0 7 .