Anxiety or negative affect in the foreign language classroom is almost universally viewed as counterproductive to language learning. This paper presents arguments that a certain amount of negative affect is necessary in order to encourage students to perform at optimal levels. Several teaching methodologies will be reviewed in terms of negative affect. A pervasive guideline, present in most modern approaches, is that negative affect must be kept as low as possible. This is perhaps most notable in the Krashen-Terrell Natural Approach paradigm, yet is also true in Suggestopedia, Counseling Learning, and TPR. Reduced affect is a commonly mentioned benefit of Computer Assisted Language Learning and other laboratory activities. Nevertheless, many methods, in spite of their professed low affect approach, employ classroom techniques which seem to raise negative affect in students. This is true of TPR, the Silent Way, and Strategic Interaction, as well as the Natural Approach. Different foreign language teaching methodologies have had very different ways of approaching the problem of affect in the classroom. Beane (1990, p. 3) defines affect as referring "to mental aspects of human nature that are differentiated from reason." He cites G. Mandler (1984) in claiming that "thought and feeling occur simultaneously in human experience." 

Stern (1984) has described classroom affect as containing three major concepts: attitude, motivation, and personality. Other researchers have established that each of these may be conceptualized as a continuum with a positive and a negative end (Spolsky, 1969; Gardner and Lambert, 1972; Gardner, 1975; Gardner, et. al., 1985; Howard, et al., 1986). 

Stern discusses three sets of affective conditions. Among these are "the affective conditions which precede the learners' approach to second language learning," and "the affective conditions that are engendered by the learning experience" (p. 385). For the purposes of applying knowledge of affective conditions to teaching, it is certainly easier to take those "engendered by the learning experience" into account, since the teacher has control over many, if not most of these. Teachers, on the other hand, generally have little control over the affective condition of students before they arrive in the classroom. 

Negative affect has been reduced in common terms to anxiety, and much research has been done on its effects in the classroom and during testing. Ruebush (1960) made an early distinction between interfering and facilitating effects of test anxiety. He wrote that "although test anxiety has interfering effects upon intellectual performance under some conditions, it may have facilitating effects under other conditions" (p. 205). Scovel (1978) has described these conditions of affect as facilitating and debilitating. Leontiev (1981), while discussing affect, makes an opposition of the terms emotional tension and operational tension as follows: 

Most language pedagogues have argued that language learning is facilitated if all negative affect is reduced; that is, language students will learn best if they have positive attitudes, high motivation, and a personality that will accept whatever particular methodological procedures are being used, and such procedures should always endeavor to reduce anxiety in the student. Gardner (1975) has summarized by stating that "more relaxed and confident students are assumed to be more proficient than those who become anxious in the language class" (from Stern, 1984, p. 384). Several teaching methodologies have been developed with an eye to reducing student anxiety, among them are Community Language Learning, Suggestopedia, and the Natural Approach (Horwitz, et al., 1986). 

Not all methodologists are in agreement, however. The reduction of all anxiety or tension is not in keeping with Leontiev's statement above that operational tension "always leads to the best possible performance." Scovel (1978) argues that a clear-cut relationship between anxiety in the classroom and foreign language achievement has not been established, and that it is premature to relate anxiety to the comprehensive task of language acquisition. 

In his review of anxiety research in foreign language learning, Scovel (1978) notes that: 

Two teaching methodologies epitomize the dichotomy that exists concerning affective considerations in the classroom: namely, the Natural Approach (Terrell, 1977; Dulay, et al., 1982) and the Silent Way (Gattegno, 1972; Richards and Rodgers, 1986). The former is based closely on the theory of language acquisition proposed by Krashen, known as the Monitor Model (Krashen, 1983, 1985, McLaughlin, 1987) which proposes an "affective filter.” This filter, according to Krashen, is raised by the student when the learning situation produces fear or anxiety, among other things. Krashen maintains that language acquisition cannot take place unless the filter, meaning any negative affect held by the student, is removed--the lower the affect, the lower the filter, and the less perturbation caused in the process of acquisition. 

One of the most notable attributes of the Natural Approach is the so-called silent period, during which the students are not required to produce in the language, lest their affective filter be raised. One of the major classroom techniques used in the Natural Approach is Total Physical Response (TPR), developed by Asher (1965, 1977). When using TPR the teacher gives commands to the students, who in turn respond physically, but generally silently to the commands. 

Proponents of the Natural Approach and the silent period claim that students should be permitted to remain silent or respond in early language instruction, in that this approximates what language learners, particularly children have been observed to do naturally (Dulay, et al., 1982, pp. 25-26). Others have responded, however, that the evidence for a natural silent period is inconclusive, and that such a period, if it exists, may be only two to three weeks rather than the one to three months claimed by Krashen (Gibbons, 1985, p. 257). 

Gattegno's Silent Way, on the other hand, does not allow for silence on the part of the student. Indeed, the silence evoked in the name of the methodology refers to silence on the part of the teacher. Among the basic techniques of the Silent Way method are the requirements that teachers model new vocabulary and structures sparingly, and that students are required to repeat and use them, and therefore produce, immediately. Gattegno wrote, "The teacher's strict avoidance of repetition forces alertness and concentration on the part of the learners" (Gattegno, 1972, p. 80). Richards and Rodgers (1986) state that the Silent Way "views learning as a problem-solving, creative, discovering activity, in which the learner is a principal actor rather than a bench-bound listener" (p. 100). 

If we are to accept today's more common treatment of affect and classroom anxiety, then the Silent Way would fail. Indeed, it seems devised to increase anxiety on the part of the student. It can be reasoned, however, that this increased anxiety results in operational tension, and thus, in improved performance and learning. Nevertheless, while insisting that students produce language early on in the learning experience, debilitating anxiety is, to a large part, avoided through the silence of the teacher. 

Lantolf (1983, p. 29) reports that a student in a Spanish course using the Silent Way commented: “Even though I personally hate to be forced to speak out in class, the participation idea in this class thus far has been better for me than regular classes.” This student’s comment is particularly important when compared to some of the data driving some researchers to conclude that all anxiety in the classroom is to be avoided. 

Horwitz, Horwitz, and Cope (1986) make this claim and base their assumption on their Foreign Language Classroom Anxiety Scale--a 33-item survey with Likert scale responses. However, all of these items ask the students how they feel in certain situations, but do not ask them what result those situations have on their ability to learn a language. For example, most students state that they would feel nervous speaking the foreign language with native speakers. It may be, however, that most of the same students would agree that doing so would help them learn the language. 

I will illustrate this with an example from TPR. In 1983, Asher argued that his method was successful because language learners were required to physically act out the commands given to them by their teachers. He cited Lieberman and Altschul’s (1971) experiment in which a series of commands were given to three groups of approximately 50 students each. The first group closed their eyes, listened to the commands and imagined performing them. The second group watched a model perform, and the third group actually performed when they heard the commands. Later, on a recall test the three groups performed in significantly different ways; most notably, the third group, the one that actually performed the commands while learning them, outperformed the other groups in recall by a large margin. Asher argues that this is evidence that the physical activity of TPR is what helps the method work. I suggest however, that it could just as plausibly be the increase in performance anxiety associated with physically responding to a teacher’s commands that results in increased recall ability. 

I propose that increasing anxiety in nonanxious or less anxious students will produce operational tension. This operational tension will stimulate negatively oriented students to perform better than they would in a low anxiety situation. Affectual pressures which approach, but do not reach, those of debilitation, create a situation in which learners forget their goals, change their orientation, and concentrate more fully on the task. 

As illustrated in Table 1, if the learners’ orientation to the task is positive, creating operational tension will certainly result in optimal work; however, if the learners’ orientation is to do poorly, increasing the tension without causing the students to freeze up, may allow them to discard plans for poor performance, and will motivate them to concentrate on responding to the best of their ability to the task at hand. 

Two recent surveys of placement testing in Spanish (Klee and Rogers, 1989, Wherritt and Cleary, 1990) mentioned the problem of false starters, defined as those students who, when taking a placement test, purposely perform poorly, so that they might begin their college studies in a beginning-level class. Neither of these studies, however, suggests a solution to this problem. These students obtain spuriously low scores and are said to fear success (Drasgow and Levine, 1985, Madsen, 1987, Griffore, 1977). Based on Leontiev's definition of operational tension and the prediction of Table 1, it is hoped that increasing affect may cause uncooperative students to perform better. 

In an early study of affect and testing, Ruebush (1960) tested 280 sixth-grade boys on an intelligence test and two anxiety scales. The boys were then divided into groups of high and low anxious, and were administered another task. Ruebush found that overall, as he had predicted, the subjects with high anxiety performed better than those with low, so long as the anxiety resulted in cautiousness rather than defensive reactions to the anxiety. 

One such defensive reaction is produced by a fear of success. Griffore (1977) concludes that fear of failure may influence performance on final exams more than fear of success, but that on placement tests, the enhanced possibility of success perceived by examinees would also enhance their fear of success. 

Galassi, et al. (1981) demonstrate that test anxiety, either fear of success or failure, is not altered through the course of a test due to assessment. That is, examinees seem to be influenced by factors relating to the perceived importance, or lack thereof, on a test, and not by knowledge of their actual performance on the test. 

Two experiments with testing and affect 

In order to test the hypothesis that a limited increase in student anxiety during a testing procedure can produce an enhancement of operational tension and better examinee cooperation, and thus result in better performance, I conducted two experiments. In each, two versions of a test were administered to different groups of students. One version was designed to lower anxiety and the other to raise it. The hypothesis to be tested predicted that the students with raised anxiety levels would outperform the nonanxiety group. 

An exercise on the use of the prepositions "por" and "para" in Spanish was administered to two second year classes at the University of Delaware in the fall semester of 1988. The activity, devised originally by Diehl (1989) and based on Labarca (1983), consisted of two passages with all occurrences of the words "por" and "para" deleted. There were 25 instances of such deletion. Students were simply required to fill in each blank with either "por" or "para", whichever they felt was most appropriate. The usage of "por" and "para" was explained in the then current chapter of their textbook (Labarca, 1988), had been discussed in class, and exercises had been completed during the previous week. I was the instructor in both sections of the course, and both had been exposed to the same amount of practice with "por" and "para". 

The design for these experiments was not strictly controlled, as the samples were not randomly selected from an entire population of Spanish students. Rather, in both experiments the samples were composed of entire Spanish classes at universities. The design can however be referred to as quasi-experimental, following Campbell and Stanley (1963). This design, called by Campbell and Stanley the Nonequivalent Control Group Design, can show both high internal and high external validity. This means, that with care, the results may be inferred to the general population. 

The two classes to which the "por" and "para" task was given were told different things regarding its purpose, in hopes that in one class there would be very little reason to worry about performing well and, therefore, little anxiety, and that in the other enough test anxiety would be created to put the students at a level of operational tension. The control group was told that they were participating in an exercise from another teacher who was gathering data for a project, and that, as a favor, they were being asked to try to fill in the passages as best they could. They were told they need not sign their names, and the activity was not timed. This was the control group. The second class was told that the exercise was a test, and that it would be counted as a test grade in the course. They were told to sign their names, and given 20 minutes to complete the activity. This was the experimental group, hereafter called the affected group. 

It was expected that the second class would outperform the first, for, given the hypothesis that increased test anxiety will cause operational tension, it is the second class that was given information to make them anxious. The first class, on the other hand, was expected to have a nearly total lack of anxiety or tension of any kind, and therefore, to perform less accurately. 

Table 2 shows the results of this "por" and "para" test. Once scored the supposed increased anxiety in the second class accounted for a mere 4.5% rise in scores. This appears on the surface to be a small amount, but the change in the affective situation, that is, telling the students it was a graded quiz, was the only change between the affected and the control groups. All other factors were constant: the rules governing the use of "por" and "para" in Spanish had been reviewed in the same way in both classes; the instructor was the same; the activity was unannounced in both classes; and neither group had any more chance or reason to study than the other. Nevertheless, the results from this experiment could not sustain the hypothesis. A t test revealed that the two classes performed in similar ways on the test. 

Fear of success may be influenced by face validity. Traditionally, the notion of face validity has favored the examiner's evaluation of a test over that of the examinee. Nevo (1985), for example, places the burden of identifying face validity on the examiner, yet Zeidner and Bensoussan (1988) suggest that: 

Nevertheless, in their study, Zeidner and Bensoussan found no significant relationship between students' attitudes and test performance. 

Lack of examinee cooperation, therefore, may be the result of a lack of examinee perceived face validity, as well as test anxiety and fear of success. 

A second experiment was performed, this time with more significant changes in the administration of the test between the two groups. I gave the same exercise to two Spanish classes at the University of Utah during the winter semester of 1991. Again, the exercise covered material that had been taught and practiced during the preceding week of class. Both classes shared the same teacher. It was decided, however, to change not only the students' perceived purpose of the exercise, as in the previous experiment, but also to change the look and length of the exercise itself. 

The one-page form of the exercise administered to the control group, that is, the class which was not expected to feel anxious or tense, was hand written and had been copied with a purple spirit master on a ditto machine. All other materials received by the students in this class previously in the semester had been typed and photocopied. There was no space provided for students to write their names. The heading on the sheet was simply "Ejercicio de gramática". The students were told, as in the former experiment, that this was a data-gathering favor for another teacher. They were not timed. 

The second class received copies of the exercise that had been typed and photocopied. With the addition of a second exercise, with no other purpose than to increase affect, the test was almost two pages long. The heading read, "Examen de grámatica", followed by a space for the student's name, the course number and semester. As was done in the previous experiment, this group was told that this was a test that would count as part of their course grade. They were allowed 20 minutes to complete it. 

The task itself was a matching exercise with examples of conjugated verbs in one column and the grammatical term naming the various possible tenses in another. The second dummy exercise was a cloze test made from a passage taken from the autobiography of St. Theresa of Jesus. The complete text of the test is in Appendix B. 

Table 3 shows the scores from both groups' performance on the test. A t test revealed the differences in the means to be significant at the < .05 level. 

In this experiment, the data seem to confirm the hypothesis that increasing affect, particularly if the subjects may be somewhat apathetic to begin with, can create the necessary operational tension to stimulate better performance. 

Conclusion 

In the two preceding experiments, affect was increased in the examinees by changing their perception of the test. In the first, the experimental group was told the test would count as a part of their course grade, and the test was timed. This alone did not cause a significant rise in their scores over the control group. In the second experiment, the experimental group was also given a more important looking test: it was longer, immaculately typed, and neatly photocopied. This, in conjunction with the students' belief that the test counted toward their course grade, resulted in significantly better performance than the control group. 

I do not necessarily propose that anxiety be produced in all language learning or language testing situations, but rather, that in those situations there is reason to believe that students may have little incentive to do well, or when some students admit that they wish to do poorly increased negative affect can cause students to function at a level of operational tension, and that at that level we can expect to obtain optimum learning and performance. 
 
 
REFERENCES 
Asher, J. J. (1977). Learning another language through actions: The complete teacher's guidebook. Los Gatos, CA: Sky Oak Productions. 

Asher, J. J., Kusudo, J. A., & De la Torre, R. (1983). Learning a second language through commands: The second field test. In J. W. Oller, Jr. & P. A. Richard-Amato (Eds.) Methods that work: A smorgasbord of ideas for language teachers. Rowley, MA: Newbury House. 

Beane, J. A. (1990). Affect in the curriculum: Toward democracy, dignity, and diversity. New York: Teacher College Press. 

Diehl, D. (1989). 'Por' y 'para': un estudio de dos preposiciones difíciles. Unpublished manuscript. 

Drasgow, F., & Levine, M. V. (1985). Optimal detection of inappropriate test scores. University of Illinois at Urbana: Model Based Measurement Lab. 

Dulay, H., Burt M., & Krashen, S. (1982). Language two. New York: Oxford University Press. 

Galassi, J. P., Fuerson, H. T., Jr., & Sharer, R. (1981). Current versus retrospective assessment in test anxiety research. Journal of Consulting and Clinical Psychology, 49, 614-615. 

Gardner, R. C. (1975). Motivational variables in second language learning. In G. Taggart (Ed.), Attitude and aptitude in second language learning: Proceedings of the Fifth Symposium of the Canadian Association of Applied Linguistics. Toronto, May 1974. 

Gardner, R. C., Lalonde, R. N., & Moorcroft, R. (1985). The role of attitudes and motivation in second language learning: Correlational and experimental considerations. Language Learning, 35, 207-227. 

Gardner, R. D., & Lambert, W. E. (1972). Attitudes and motivation in second language learning. Rowley, MA: Newbury House. 

Gattegno, C. (1972). Teaching foreign languages in the schools: The silent way. New York: Educational Solutions. 

Gibbons, J. (1985). The silent period: An examination. Language Learning, 35, 225-267. 

Griffore, R. J. (1977). Fear of success and task difficulty: Effects on graduate students' final exam performance. Journal of Educational Psychology, 69, 556-563. 

Horwitz, E. K., Horwitz, B. H., & Cope, J. (1986). Foreign language classroom anxiety. Modern Language Journal, 70, 125-132. 

Howard, D., Kelsey, G., & Mathews, T. J. (1986). The relationship between language play and motivation in foreign language and second language learners. Unpublished manuscript. [cf. Lantolf, et al., 1987] 

Klee, C., & Rogers, E. (1989). Status of articulation: Placement, advanced placement, credit and course options. Hispania 72, 763-771. 

Krashen, S. (1983). The input hypothesis. In Oller (Ed.). Issues in language testing research. Rowley, MA: Newbury House. 

Krashen, S. (1985). The input hypothesis: Issues and implications. London: Longman. 

Labarca, A. (1983). Invitación a escribir. [Workbook to accompany Invitación] New York: Holt, Rinehart and Winston. 

Labarca, A., & Hendrickson, J. M. (1988). Nuevas dimensiones. Boston: Heinle and Heinle. 

Lantolf, J. P. (1983). Silent way in the university setting. Unpublished manuscript. 

Lantolf, J. P., Mathews, T. J., Kelsey, G., & Howard, D. (1987). Language play and adult language learning. Paper presented at the Second International Symposium on Applied Psycholinguistics, Kassel, Federal Republic of Germany. 

Leontiev, A. A. (1981). Psychology and the language learning process. Oxford: Pergamon Press. 

Lieberman, L. R., & Altschul, S. (1971). Memory for a list of commands: Imagining, seeing, doing. Perceptual and Motor Skills, 33, 530. 

Madsen, H. S. (1987) Utilizing Rasch analysis to detect cheating on language examinations. In K. M. Bailey, et al. (Eds.), Language testing research: Selected papers from the Colloquium. Monterey, CA. February 27-28, 1986. 

Mandler, G. (1984). Mind and body: Psychology of emotion and stress. New York: Norton. 

McLaughlin, B. (1987). Theories of second-language learning. London: Edward Arnold. 

Nevo, B. (1985). Face validity revisited. Journal of Educational Measurement, 22, 287-293. 

Richards, J. C., & Rodgers, T. S. (1986). Approaches and methods in language teaching. Cambridge: Cambridge University Press. 

Ruebush, B. K. (1960). Interfering and facilitating effects of test anxiety. Journal of Abnormal and Social Psychology, 60, 205-212. 

Scovel, T. (1978). The effect of affect on foreign language learning; A review of the anxiety research. Language Learning, 28, 129-142. 

Spolsky, B. (1969). Attitudinal aspects of second language learning. Language Learning, 19, 272-283. 

Stern, H. H. (1984). Fundamental concepts of language teaching. Oxford: Oxford University Press. 

Terrell, T. (1977). A natural approach to second language acquisition and learning. Modern Language Journal, 61, 325-337. 

Wherritt, I., Cleary, T. A. (1990). A national survey of Spanish placement or outcome assessment at BA-granting institutions in the United States. Foreign Language Annals 23, 157-165. 

Zeidner, M., & Bensoussan, M. (1988). College students' attitudes towards written versus oral tests of English as a foreign language. Language Testing, 5, 100-114.