includeLANGUAGE ANXIETY AND ACHIEVEMENT

Annual Review of Applied Linguistics

(2001) 21, 112-126. Printed in the USA. Copyright ?

2001 Cambridge University Press 0267-1905/01 $$9.50

7. LANGUAGE ANXIETY AND ACHIEVEMENT

Elaine K. Horwitz

This chapter considers the literature on language learning anxiety in an effort to clarify

the relationship between anxiety and second language learning. It will first argue that

language anxiety is a specific anxiety

—

rather than a trait anxiety

—

and discuss how

this conceptualization has helped clarify the research literature. After Horwitz,

Horwitz, and Cope (1986) proposed that a specific anxiety construct which they called

Foreign Language Anxiety was responsible for students’ uncomfortable experiences

in language classes and offered an instrument, the Foreign Language Classroom

Anxiety Scale (FLCAS), to measure this anxiety, findings concerning anxiety and

language achievement have been relatively uniform, indicating a consistent moderate

negative relationship between anxiety and achievement. However, some researchers

(Sparks and Ganschow and their colleagues) have suggested that poor language

learning is a cause rather than a result of language anxiety. This review concludes that

anxiety is indeed a cause of poor language learning in some individuals and discusses

possible sources of this anxiety, including difficulty in authentic self- presentation and

various language teaching practices. In addition, it reports on new trends in language

anxiety research that attempt to identify aspects of language learning (e.g., reading

anxiety or writing anxiety) which provoke anxiety for some individuals.

Researchers, language teachers, and even language learners themselves have been

interested in the possibility that anxiety inhibits language learning for quite some time.

Clinical experience, empirical findings, and personal reports all attest to the existence of

anxiety reactions with respect to language learning in some individuals; however, the

research history in this area has not been straightforward. This chapter considers the

literature on language learning anxiety in an effort to clarify the relationship between

anxiety and achievement in second language learning.

Anxiety as Psychological Construct

“Anxiety is the subjective feeling of tension, apprehension, nervousness

, and

worry associated with an arousal of the autonomic nervous system,” (Spielberger, 1983, p.

112

LANGUAGE ANXIETY AND ACHIEVEMENT

113

1). Not only is it intuitive to many people that anxiety negatively influences language

learning, it is logical because anxiety has been found to interfere with many types of

learning and has been one of the most highly examined variables in all of psychology and

education. Psychologists distinguish several categories of anxiety. Typically, anxiety as a

personality “trait” is differentiated from a transient anxiety “state.” In other words, trait

anxiety is conceptualized as a relatively stable personality characteristic while state anxiety

is seen as a response to a particular anxiety- provoking stimulus such as an important test

(Spielberger, 1983). More recently the term situation-specific anxiety has been used to

emphasize the persistent and multi- faceted nature of some anxieties (MacIntyre & Gardner,

1991a). Public speaking anxiety is generally viewed to be in this category, and this chapter

will take the position that foreign language anxiety is as well.

Early Perspectives on Anxiety and Language Learning

Since the mid 1960s scholars have entertained the possibility that anxiety

interferes with second language learning and performance; however, documentation of that

relationship came much later. Interestingly, the relationship between anxiety and second

language achievement

—

the subject of this review

—

is exactly the same issue that puzzled

Scovel over two decades ago (Scovel, 1978). Scovel reviewed the then available literature

on anxiety and language learning in an attempt to explain a truly conflicting set of findings.

At the time there were studies which found the anticipated negative relationship between

anxiety and second language achievement, but several studies found no relationship, and

positive relationships between anxiety and second language achievement were also

identified (Chastain, 1975; Kleinmann, 1977). In other words, contrary to the predictions of

many language teachers, some studies found that learners with higher levels of anxiety

actually showed higher achievement scores. Scovel posited a rational solution to this

enigma. He argued that since the various studies used different anxiety measures such as

test-anxiety, facilitating- debilitating anxiety, etc., they logically found different types of

relationships between anxiety and language achievement. Scovel concluded that language

researchers should be specific about the type of anxiety they are measuring and

recommended that anxiety studies take note of the myriad of types of anxiety that had been

identified.

Language Anxiety

Scovel’s suggestions have proven to be good ones, and since that time researchers

have been careful to specify the type of anxiety they are measuring. However, in 1986,

Horwitz, Horwitz, and Cope took the literature a step further by proposing that a

situation-specific anxiety construct which they called Foreign Language Anxiety was

responsible for students’ negative emotional reactions to language learning. According to

Horwitz, Horwitz, and Cope, this anxiety stems from the inherent inauthenticity associated

with immature second language communicative abilities:

Adults typically perceive themselves as reasonably intelligent,

socially-adept individuals, sensitive to different socio-cultural mores.

These assumptions are rarely challenged when communicating in a

native language as it is not usually difficult to understand others or to

114 ELAINE K. HORWITZ

make oneself understood. However, the situation when learning a

foreign language stands in

marked contrast. As an individual’s

communication attempts will be evaluated according to uncertain or

even unknown linguistic and socio- cultural standards, second language

communication entails risk-taking and is necessarily problematic.

Because complex and nonspontaneous mental operations are required

in order to communicate at all, any performance in the L2 is likely to

challenge an individual’s self

-concept as a competent communicator

and lead to reticence, self-consciousness, fear, or even panic (p. 128).

They also offered an instrument, the Foreign Language Classroom Anxiety Scale

(FLCAS), to measure this anxiety. Since that time, findings concerning anxiety and

language achievement have been relatively uniform. Studies using the FLCAS and other

specific measures of second language anxiety have found a consistent moderate negative

correlation between the FLCAS and measures of second language achievement (typically

final grades). Accordingly, this review will be limited to those studies which assume a

specific anxiety reaction to language learning. In addition to the FLCAS, these measures

include the French Class Anxiety Scale (Gardner & Smythe, 1975), The English Use

Anxiety Scale (Clement, Gardner, & Smythe, 1977), the English Test Anxiety Scale

(Clement, Gardner, & Smythe, 1980), the French Use Anxiety Scale (Gardner, Smythe, &

Clement, 1979), and the Spanish Use Anxiety Scale (Muchnick & Wolfe, 1982).

The situation-specific anxiety in response to language learning proposed by

Horwitz, Horwitz, and Cope has been found to be largely independent of other types of

anxiety. Horwitz (1986) found a low but significant correlation between the FLCAS and

Spielberger’s (1983) test of trait

-anxiety (r = .29

, p =

.002). Slight positive, but

nonsignificant, correlations were also found between the FLCAS and Fear of Negative

Evaluation (Watson & Friend, 1969) and Communication Apprehension (McCroskey,

1970). In the case of test anxiety (Sarason, 1978), the correlation was moderate and

significant (r = .53,

p =

.001). Even so, this correlation means that the two measures only

share 28% of variance and are, therefore, reasonably independent. Also consistent with the

construct of a

situation- specific

anxiety,

MacIntyre

and

Gardner

(1989)

did

not

find

a

relationship

between general anxiety and foreign language vocabulary learning.

Thus, with the development of distinct situation-specific measures of foreign

language anxiety, the issue of appropriate anxiety measurement seemed to be resolved;

however, the issue of appropriate outcome measures remained. Steinberg and Horwitz

(1986) argued that the use of final grades as a measure of second language achievement was

probably in and of itself a source of variability in the anxiety literature and urged

researchers to use more subtle achievement measures to capture the true effects of anxiety.

Indeed, this study found that ESL students attempted a greater number of elaborated and

personal messages in English when experiencing an experimental condition intended to

relax them than those learners experiencing a treatment designed to induce anxiety.

Importantly, these differences in elaboration and number of personal utterances were

observed even though the anxious and nonanxious students displayed equal levels of

overall oral fluency.

LANGUAGE ANXIETY AND ACHIEVEMENT

115

Language Anxiety and Achievement

There have been a number of studies in a number of instructional contexts with

varying target languages which find a negative relationship between specific measures of

language anxiety and language achievement. In the first study using the FLCAS (Horwitz,

1986), there was a significant moderate negative correlation between foreign language

anxiety and the grades students expected in their first semester language class as well as

their actual final grades, indicating that students with higher levels of foreign language

anxiety both expected and received lower grades than their less anxious counterparts.

MacIntyre and Gardner (1989) also found significant negative correlations between a

specific measure of language anxiety (French class anxiety) and performance on a

vocabulary learning task.

With respect to a target language which is typically perceived as difficult by

English-speakers, Aida (1994) found a significant negative correlation between FLCAS

scores and final grades among American second-year Japanese students. This finding was

replicated by Saito and Samimy (1996) with Japanese learners at three levels (beginning,

intermediate, and advanced). Similarly, in a study of Canadian university learners of French,

Coulombe (2000) found a somewhat smaller (but significant) negative correlation between

FLCAS scores and final grades in eleven French classes ranging from beginning to

advanced. Thus, it appears that the observed negative relationship between anxiety and

achievement holds at various instructional levels as well as with different target languages.

However, it should also be noted that all the studies reviewed here included students at the

college or university level and the relationship between anxiety and achievement in

younger learners remains relatively unexplored (see, however, Sparks & Ganschow, 1996).

One study is particularly interesting because it focuses on more advanced

language learners (pre-service teachers) in a non-North American context. Rodriguez (1995)

found a significant negative correlation between FLCAS scores and final grades among

Spanish-speaking EFL students in seven English classes in Venezuela. In this case the

correlation (r = -.57) was somewhat higher than reported in the other studies, indicating that

the two measures have almost one-third of the variance in common. Considering all the

possible influences on final grades, this seems a very substantial correlation and raises the

possibility that language anxiety is also an important issue among language teachers, as

Horwitz (1996) argues.

Finally, in an Asian EFL context, Kim (1998) not only found significant negative

relationships between FLCAS scores and final grades but also reported an interesting

difference in the relationship when observed in a traditional reading- focused class and a

conversation class. Specifically, students were considerably less anxious in the reading

class than in the conversation class. Thus, this study appears to support teachers’ and

students’ intuitive feelings that

language classrooms which require oral communication are

more anxiety-provoking than traditional classrooms.

Several studies have also noted a negative relationship between language anxiety

and outcome measures other than final grades. Try long (1987) found a negative

relationship between anxiety and teacher ratings of achievement; MacIntyre, Noels, and

Clement (1997) observed a negative relationship between anxiety and students’ self

-ratings

116 ELAINE K. HORWITZ

of their language proficiency. Gardner and MacIntyre (1993) present perhaps the most

extensive set of findings with respect to language anxiety. Using measures of both

classroom anxiety and language use anxiety, they found significant negative correlations

with several language production measures including a cloze test, a composition task, and

an objective French proficiency measure. Interestingly, they found somewhat higher

negative correlations between student anxiety scores and their self-ratings of French

competence than with their actual performance on the tests of French ability. Finally, with

respect to some of the nonlinguistic but hoped-for goals of language instruction, Spitalli

(2000) found a negative relationship between FLCAS scores and a measure of attitudes

toward people of different cultures in American high school language learners of French,

Spanish, and German.

One final point about the achievement studies seems relevant; that is, levels of

foreign language anxiety may vary in different cultural groups. While Horwitz (1986) and

Aida (1994) found relatively similar means on the FLCAS for American foreign language

learners, Truitt (1995) found relatively higher levels in Korean EFL learners, and Kunt

(1997) found somewhat lower levels in Turkish and Turkish-Cypriot learners of English. It

also seems, as reported above, that for American learners of foreign languages, at least,

anxiety levels do not vary with respect to target language (Aida, 1994; Horwitz, 1986; Saito,

Horwitz, & Garza, 1999).

Possible Confounding in the Achievement Studies

Whenever the statistical procedure of correlation is used, as it was in the

achievement studies reviewed here, it is not possible to be sure of the direction of the

correlation or to rule out the possibility that some uncontrolled variable is responsible for

any relationship which has been observed between the two variables under study. In the

case of anxiety, it seems especially important to consider that students who do poorly in

language classes would naturally (and logically) become anxious. These concerns form the

basis of a number of papers which seek to determine sources of foreign language anxiety.

For example, Young (1986) found that a significant correlation between anxiety and

performance on an oral examination in French and Spanish dissipated to a chance level

when actual second language ability was controlled. She reminded readers, however, that

the oral examination used in the study was only a practice one and that the participants were

advanced language learners (pre-service teachers). She therefore speculated that these

students were not experiencing sufficient levels of anxiety to inhibit their performance. In

addition, she suggests that in the case of advanced learners such as these, anxiety would

probably be a greater hindrance to their ability to perform than to their development of L2

proficiency. (See also Rodriguez, 1995, and Horwitz, 1996).

Contrary to Young’s results, Phillips (1992) found a significant negative

correlation between FLCAS scores and performance on an oral interview examination even

when ability in the form of students’ written examination averages was statistically

controlled. This study may be more representative of the relationship between anxiety and

oral performance in actual language classes than Young’s because the students

were

participating in an oral interview for which they would receive a grade and were

third-semester college French students rather than pre-service teachers.

LANGUAGE ANXIETY AND ACHIEVEMENT

117

There has also been a strand of research which has strongly questioned the

existence of foreign language anxiety independent of language achievement. In a series of

studies, Sparks and Ganschow and their colleagues (see for example, Ganschow, Sparks,

Anderson, Javorsky, Skinner, & Patton, 1994; Sparks & Ganschow, 1991; Sparks &

Ganschow, 1996, this volume) propose the Linguistic Coding Differences Hypotheses

(LCDH) and offer a theoretical perspective for understanding relationships between foreign

language anxiety and foreign language achievement: in their view, because

FL (foreign language) learning i

s based primarily on one’s native

language learning ability (i.e., language aptitude), students’ anxiety

about FL learning is likely to be a consequence of their FL learning

difficulties, and students’ language learning ability is a confounding

variable when studying the impact of affective differences (e.g., anxiety,

motivation, attitude) on FL learning (Sparks, Ganschow, & Javorsky,

2000, p. 251).

Thus, Sparks and Ganschow pose a crucial question with respect to the study of foreign

language anxiety and its relationship with achievement. Specifically, they ask if anxiety is a

cause or result of poor achievement in language learning and propose subtle first language

learning deficits as the primary cause of poor achievement. Language difficulties, they

prop

ose, “are likely to be based in native language learning and that facility with one’s

language “codes” (phonological/ orthographic, syntactic, semantic) is likely to play an

important causal role in learning a FL” (p. 235). Indeed, Ganschow and Sparks (199

1)

found that less anxious language learners performed significantly better on oral and written

foreign language measures as well as on the Modern Language Aptitude Test. (See also

Sparks and Ganschow, this volume). It is easy to conceptualize foreign language anxiety as

a result of poor language learning ability. A student does poorly in language learning and

consequently feels anxious about his/her language class. Conversely, a student might do

well in the class and feel very confident. The challenge is to determine the extent to which

anxiety is a cause rather than a result of poor language learning.

MacIntyre (1995a, 1995b) and Horwitz (2000) have responded to the LCDH

arguing for the existence of language anxiety independent of first or general language

learning disabilities. Particularly, they insist that anxiety is a well-known source of

interference in all kinds of learning and wonder why the case of language learning should be

different. In addition, the numbers of people who experience foreign language anxiety

appear to be far greater than the incidence of decoding disabilities in the general population,

and many successful language learners also experience language anxiety. Perhaps most

importantly, they observe that language learning requires much more than sound-symbol

correspondences and argue that the LCDH is ultimately based on an overly simplified view

of language learning.

From all these perspectives, it appears that language anxiety fits the general criterion for an

anxiety which by definition is an unrealistic reaction to a particular situation. Anxious

language learners feel uncomfortable with their abilities even if their objective abilities are

good.

Anxiety under Different Instructional Conditions