Predictors of Foreign Language Anxiety: A Study of College-Level L2 Learners of Chinese

2.1 Motivation

Scholars and researchers believe that anxiety and motivation are two theoretical constructs closely related to each other. For example, anxiety has formed a component of the Attitudes/Motivation Test Battery (Gardner, 1985) in the framework of the socio-educational model. Clément (1980, 1986) viewed anxiety and motivation as two closely related learner characteristics in his social-context model. Phillips (1992) concluded that anxiety could have a negative impact on students’ attitudes and motivation toward language study. Liu and Huang (2011) found a significant negative correlation between anxiety and language learning motivation among EFL learners. However, Tóth (2007) examined the relationship between motivation and Foreign Language Anxiety quantitatively and found that the correlation between motivation and Foreign Language Anxiety did not reach a significant level. As can be seen, the relationship between motivation and anxiety needs further clarification.

2.2 Age

Many language learners intuitively believe that younger learners are better at learning foreign languages than their adult counterparts (Horwitz, 1988). This intuition seems to be justified to some extent as some research has shown that younger people tend to be able to speak a foreign language with less accent and have stronger ability in mastering the finer points of language such as phonology and morphology (Lieberman, 1984; Newport, 1986). Studies on age and test anxiety have generally reported a positive relationship between age and test anxiety (Crook, 1979; Hunt, 1989), indicating older language learners are more anxious over taking tests. In a study exploring 26 associated variables of anxiety, Onwuegbuzie, Bailey, and Daley (1999) found that age was a significant predictor of Foreign Language Anxiety and that older students had higher levels of anxiety. It would be interesting to see whether age affects the anxiety levels of L2 learners of Chinese.

2.3 Self-perceived Achievement

Many studies have investigated the relationship between Foreign Language Anxiety and second/foreign language achievement. These studies have generally reported a consistent moderate negative relationship between language anxiety and language achievement (Horwitz, 2001). Researchers have most commonly used final course grades and standardized proficiency tests to measure language achievement in language anxiety studies (Aida, 1994; Horwitz, 1986; MacIntyre & Gardner, 1994; Phillips, 1992). However, many studies have shown that learners’ self-perceived achievement is a better predictor of Foreign Language Anxiety than actual achievement based on course grades or objective proficiency measures (Cheng, 2002; Gardner & MacIntyre, 1993b; Tóth, 2007). For example, MacIntyre, Noels, and Clément (1997) asked students to rate their language proficiency levels themselves and discovered a negative relationship between anxiety and students’ self-ratings. Onwuegbuzie, Bailey, and Daley (1999) found expected overall average for the language courses to be the best predictor of Foreign Language Anxiety among 26 potential predictors. An investigation of the relationship between anxiety and self-perceived achievement among college-level L2 learners of Chinese in the United States may add to the existing body of knowledge on this issue.

2.4 Perception of the Difficulty of the Target Language

Some studies seem to indicate that the target language has an effect on Foreign Language Anxiety (Luo, 2012). For example, compared to the anxiety levels of learners of other languages in other studies, Chinese language learners in Chinese study-abroad programs in Le’s (2004) study experienced substantially higher levels of anxiety, perhaps due to the difficulty level of Chinese. Saito, Garza, and Horwitz (1999) found that levels of foreign language reading anxiety were significantly different for the three target languages (Japanese, Russian and French) examined in their study. When reading, learners of Japanese were the most anxious, followed by the French participants, with the Russian participants experiencing the lowest levels of reading anxiety. The researchers argued that “reading Japanese provoked higher anxiety levels than reading Russian or French was anticipated due to the unfamiliar and non-Roman writing system as well as the foreign cultural content” (p. 212). In Aida’s (1994) study, the anxiety level of students of Japanese was slightly higher than that of Horwitz’s (1986) study of Spanish language learners. According to Aida, this result was understandable because “students may feel more anxious in learning a non-western, foreign language like Japanese than in learning commonly taught Western languages such as Spanish” (p. 158). However, the contribution of perception of the difficulty of the target language to Foreign Language Anxiety has hardly been investigated. Chinese is a relatively difficult foreign language for L2 learners due to its tonal and non-alphabetic nature. An exploration of the contribution of perception of the difficulty of the Chinese language to anxiety may provide new insight to this issue.

2.5 Self-perceived Language Learning Ability

Foreign language aptitude or learning ability as a potential factor contributing to Foreign Language Anxiety was proposed by Sparks and Ganschow (Ganschow & Sparks; 1996; Sparks & Ganschow, 1995). Ganschow and Sparks (1996) reported a medium size negative correlation (r = -.43, p = .008) between learners’ Foreign Language Anxiety and language aptitude as measured by the Modern Language Aptitude Test. However, Tóth (2007) found foreign language aptitude not to be significantly correlated with anxiety among first-year Hungarian English majors. Compared to foreign language learning ability based on objective measures, self-perceptions of language learning seem to be a more established predictor of Foreign Language Anxiety. Ganschow and Sparks (1991) found that students’ perception of the ease of learning a foreign language to be a strong identifier of foreign language anxiety. According to Ganschow et al. (1994), students with high levels of anxiety perceived their language course to be difficult, whereas students with low levels of anxiety perceived their language course to be easy. Horwitz (1990, 1995) reported that highly anxious learners were more likely to consider language learning a difficult task than their low-anxiety counterparts. However, very few studies have directly examined the relationship between self-perceived language learning ability and Foreign Language Anxiety. Thus, a quantitative study examining self-perceived foreign language learning ability in relation to anxiety among L2 learners of Chinese seems to be interesting.

2.6 Perceived Importance of the Target Language and Self-expectation

The last two selected learner variables, namely, perception of the importance of the target language and self-expectation in the foreign language class, have not received much attention in anxiety research to date. These two variables were identified from Luo’s (2011) interview study with Chinese language learners and teachers, who reported perception of the importance of the Chinese language and self-expectation in learning Chinese to be contributing factors of anxiety in Chinese classes. More specifically, according to the Chinese language learners and teachers being interviewed, students who viewed Chinese as of greater importance to them and who had higher self-expectation in learning Chinese, had a greater propensity to experience high levels of anxiety in learning Chinese. However, this view has not been investigated quantitatively. It is believed that a quantitative study of the relationship between the two variables and anxiety may be necessary.

4.1 Participants

Altogether, 428 (225 males, 202 females; 1 missing the gender indication) Chinese language learners at two large public universities in the U.S participated in this study and turned in valid responses. Among all the participants, 171 learners were identified as heritage learners with home exposure of Chinese, whose responses were then excluded from data analysis in this study. Therefore, participants of this study were the remaining 257 (147 males, 110 females) L2 learners of Chinese with an age range of 15 to 59 (M = 21.3, SD = 4.7) from the two universities, with 112 participants from one university and 145 from the other. They were all taking credit-bearing Chinese language courses at the two universities: 128 took Chinese as an elective and 129 took it to fulfill requirements.

Among the 257 participants, 141 were taking first-year Chinese, 76 were in second-year Chinese classes, and 40 were third-year Chinese language students. In terms of ethnicity, 186 of the participants were white, 4 were Chinese Americans with no home exposure of Chinese, 26 were Asian but not Chinese Americans, 19 were Asian international students, 16 were Hispanic, 1 was African American, and 5 were from other ethnic backgrounds.

4.2 Instruments

The instruments for this study included the Chinese Language Learning Anxiety Scale (Luo, 2014a) and a Background Questionnaire. The Chinese Language Learning Anxiety Scale is a 16-item self-report measure that consists of items scored on a 5-point Likert Scale, ranging from strongly agree to strongly disagree. In this scale, items were evenly distributed for anxieties related to the four skills, that is, speaking, listening, reading, and writing. This anxiety scale was reported to have high reliability and validity, with the internal consistency reliability (using Cronbach’s Alpha) achieving .90 (Luo, 2014a). Chinese language learning anxiety was calculated as the sum of the number of items in the scales. The possible range of scores for Chinese language learning anxiety was 16-80.

In addition to background information such as gender, age, ethnicity, college year, proficiency level, native language, and heritage-learning status, the Background Questionnaire asked the participants to estimate a grade (i.e., A, A-, B+, B-, ..., F) they expected to get in the Chinese class, and provide a rating of motivation in learning to speak, comprehend, read, and write Chinese, self-perceived foreign language learning ability, the level of self-expectation in learning Chinese, perception of the importance of the Chinese language, and perception of the difficulty level of the Chinese language on a 5-point Likert Scale.

The possible range for “self-perceived achievement” (i.e., expected grades) was 1-7, with 1 corresponding to F and 7 corresponding to A. Participants’ motivation in learning Chinese is calculated as the sum of ratings on motivation associated with the four skills. Thus, the possible range of scores for “motivation” was 4-20. The four variables, that is, perception of the difficulty of the Chinese language, self-perceived language learning ability, perception of the importance of the Chinese language, and self-expectation in learning Chinese, were each represented by one item on a 5-point Likert Scale. Therefore, the possible ranges for these four variables were all 1-5.

4.3 Data Analysis Methods

To investigate the relationship between anxiety and the seven selected learner variables, Pearson correlations and multiple regression analysis were used. Pearson correlations were computed to calculate the strength and direction of the relationship between anxiety and each of the seven variables. Multiple regression analysis was conducted to assess the effect of seven correlates simultaneously, to determine their relative contribution to the prediction of anxiety, and to identify the best predictors.

Before multiple regression analysis was computed, the researcher checked Cook’s distance and Leverage values for outliers, examined the P-P plot for normality of residuals, plotted the standardized residuals against the standardized predicted values to check linearity and equality of variances (See Figure 1) and studied the correlation matrix of all the independent variables for multicollinearity. The maximum value of Cook’s distance was .004 < 4/n, i.e., 4/257 = .016, and the maximum value of Leverage was .027 < (2k+2)/n, i.e., (2^*7+2/257) = .062, indicating no obvious outliers. A close examination of the correlation matrix showed that none of the correlations had a Pearson’s coefficient of 0.80 or greater. Thus, all the results indicated that multiple regression analysis was appropriate for the current data.

Figure 1

Residual plots

5.1 Correlation Findings for Research Question 1

The descriptive statistics of the seven variables are outlined in Table 1. Table 2 presents the correlations between each of the seven selected variables and Chinese as a Foreign Language (CFL) Anxiety.

It can be seen from Table 2 that Chinese language learning anxiety correlated significantly with all the seven variables at the 0.01 level. Chinese language learners’ anxiety was found to have positive correlations with two variables: perception of difficulty of Chinese (r = .398) and age (r =.193), and negative correlations with the remaining five variables, namely, self-perceived achievement (r = -.390), motivation in learning Chinese (r = -.317), self-perceived language learning ability (r = -.356), self-expectation in learning Chinese (r = -.246) and perception of the importance of Chinese (r = -.214).

To put it another way, older students and students who perceived Chinese to be more difficult were more anxious; students who expected to get a higher grade, who were more motivated in learning Chinese, who perceived themselves to be better at learning languages, who had a higher expectation in learning Chinese, and who perceived Chinese to be of greater importance, tended to experience less anxiety in Chinese classes.

Perception of the difficulty of the Chinese language had by far the largest correlation (r = .398) with Chinese language learning anxiety, which, by itself, explained 15.8% of the variance. The second largest correlate of anxiety was self-perceived achievement (r = -.390), accounting for 15.2% of the variance. The strength of correlation between self-perceived language learning ability and anxiety (r = -.356) ranked the third, followed by motivation, self-expectation in learning Chinese, and perception of the importance of Chinese, with age having the smallest correlation with anxiety.

In order to provide a more detailed explanation for the relationship between age and Chinese language learning anxiety, the age range and the frequency of each age group of the sample were examined. The ages of the sample ranged from 15 to 59. The majority of the participants (n = 239) aged between 18 and 29, with only 14 participants being 30 or above. Figure 2 shows the Scatter Plot of age and Chinese language learning anxiety. Based on the pattern shown in Figure 2, it takes caution to interpret the small negative correlation between age and anxiety. If only participants aged between 18 and 29 were studied, the correlation analysis between age and anxiety may have produced different results.

Figure 2

Scatter plot between age and CFL anxiety

5.2 Multiple Regression Findings for Research Question 2

This study used multiple regression analysis to examine the simultaneous effect of the seven selected variables on Chinese Language Learning Anxiety. More specifically, this procedure was used to determine the relative contribution of these variables to the anxiety level of L2 learners of Chinese and to identify the best predictors of Chinese Language Learning Anxiety. Table 3 presents the standardized regression coefficients, t value, and squared partial correlation coefficients of the initial model, with all seven learner variables entered in the regression equation. The table shows the more parsimonious model suggested by the backward selection method.

The results in Table 3 shows that the seven variables together contributed significantly, F(7, 254) = 19.06, p = .0001, to the prediction of Chinese Language Learning Anxiety. The R value (.592) for the initial model indicates a moderate-strong linear relationship between the seven learner variables and Chinese Language Learning Anxiety.

The seven variables combined to explain 35.1% of the variance in Chinese Language Learning Anxiety, which indicates a large effect size according to Cohen’s (1988) criteria for determining the effect of a set of independent variables. The Adjusted R Square indicating the estimated coefficient for the population was 33.2%.

Moreover, five variables (i.e., motivation, age, self-perceived language learning ability, self-perceived achievement, and perception of the difficulty of the Chinese language) had significant contribution to the prediction of Chinese language learning anxiety at the 0.01 level, whereas two variables (i.e., perception of the importance of Chinese and self-expectation in learning Chinese) were not significant predictors (p = .259, .778).

The squared partial correlation coefficients were often recommended to assess the relative contribution of individual variables in multiple regression analysis (e.g., Cohen, 1988). According to the squared partial correlation coefficients, perception of the difficulty of the Chinese language was the best predictor of Chinese Language Learning Anxiety, explaining 9.99% of the variance on its own. The unique contribution of self-perceived achievement was the second largest, accounting for 7.29% of the variance, followed by self-perceived language learning ability and age, with motivation explaining the least amount of variance among all the significant predictors, explaining 4.37%, 3.46%, and 2.50% of the variance in anxiety respectively.

Interestingly, the two variables, perception of the importance of Chinese and self-expectation in learning Chinese, reported to correlate significantly with anxiety in the correlation analysis, were found not to be significant predictors in the multiple regression analysis when all the seven variables were examined simultaneously. In order to seek explanations for this result, Table 4 presents the correlation matrix of the seven variables.

As the matrix shows, self-expectation in learning Chinese had a moderate significant correlation with the perception of the importance of Chinese (r = .57, p < .01). In addition, self-expectation in learning Chinese was also found to significantly correlate with motivation (r = .53), self-perceived language learning ability (r = .43), and self-perceived achievement (r = .38) at the 0.01 level. Similarly, perception of the importance of Chinese was reported to have a significant moderate correlation with motivation (r = .62) and significant slight correlations with self-perceived language learning ability (r = .29) and self-perceived achievement (r = .19) at the 0.01 level. Although these correlations are not high individually, moderate correlations among independent variables may still confound the effect of predictors in multiple regression (Stevens, 2009). This may partly explain the above mentioned result related to perception of the importance of Chinese and selfexpectation in learning Chinese.

Since perception of the importance of Chinese and self-expectation in learning Chinese were found not to be significant predictors of Chinese Language Learning Anxiety, they were removed from the list of independent variables in the final model. The results of Multiple Regression analysis of this model were reported in Table 5. After the two variables were removed, there was only a slight drop for R Square (.351 vs. .347), indicating that perception of the importance of Chinese and self-expectation in learning Chinese only accounted for 0.4% of the variance in Chinese Language Learning Anxiety. The five variables in the final model explained 34.7% of the variance in Chinese Language Learning Anxiety.

Adjusted R Square for this model was slightly larger than for the initial model (.334 vs. .332), and the standard error decreased (9.49 vs. 9.47), indicating “a better match between the predicted and the observed values of participants’ anxiety scores” (Tóth, 2007, pp. 136-137).

In this model, the two best predictors of Chinese Language Learning Anxiety remained perception of the difficulty of the Chinese language and self-perceived achievement, followed by self-perceived language learning ability, age, and motivation.

The final multiple regression model shows that students with the highest level of anxiety tended to have one of the following characteristics: perceiving the Chinese language to be difficult, expecting to get a high grade in Chinese class, perceiving themselves to be poor at foreign language learning, being older, or having low motivation in learning Chinese.