Brian Asbjornson

Critical to the study of the influence of beliefs on language learning is a basic definition of beliefs. Kuntz (1997) defines beliefs as “notions about language learning that students have acquired” (p.4), whereas Banya and Cheng (1997) define beliefs as an individual’s opinions about language learning. Indeed, beliefs are often considered to be synonymous with attitudes, representations, opinions or ideologies (Fraser & Gaskell, 1990). These definitions, however, do not define the construct of language learning beliefs beyond a very rudimentary level.

Gaies (1998) proposes three characteristics of beliefs about language and language learning: 1) beliefs are subjective understandings; 2) beliefs are relatively stable and 3) beliefs are idiosyncratic. First, beliefs are subjective, therefore, they do not necessarily reflect external reality. Beliefs reflect “truths” held by individuals and, therefore, are potentially facilitative for language learning (e.g., everyone can learn a language) or debilitative (e.g., boys are not good language learners). Second, beliefs about language learning should be stable over time, which indicates that beliefs are formed and maintained by a complex social and cultural system. Finally, Gaies asserts that belief systems are unique to individuals.

Several models have proposed that beliefs are an important factor in language learning achievement. Following is a critique of two models, and then a new model is proposed.

Abraham and Vann (1987) propose that a learner’s philosophy, which they define as beliefs about how language is learned, determines the approach and subsequent strategies a learner takes when learning a second language. It is the combination of philosophy, approach and strategies that determine success or failure at language learning. Abraham and Vann posit that background factors such as intelligence, personality, education and cognitive style, along with the environmental factors of formal and informal instruction determine a learner’s philosophy.

The Abraham and Vann model, however, is problematic for several reasons. First, the model does not include anxiety, motivation and attitude variables—all variables that have been shown to be important factors in language learning (for example, Gardner, Tremblay & Masgoret, 1997). Moreover, the model does not include an aptitude component, although it is possible that they have included aptitude as part of intelligence. These problems are perhaps because the Abraham and Vann model is a conceptual model and was not based on quantitative data.

A more comprehensive model of how beliefs might be an important factor in language achievement comes from an empirical study done in China (Wen & Johnson, 1997). In their study Wen and Johnson used measures of L1 and L2 proficiency and several beliefs and strategies measures. The result was a model that showed a causal, direct relationship between sex, L1 proficiency, vocabulary learning strategy, L2 proficiency and L2 achievement. The indirect factors were the learners’ strategies, but these strategies were directly related to the beliefs about the strategies. Beliefs are likely to be an underlying factor in the use of strategies.

Wen and Johnson see language achievement as the product and modifiable learner factors of learning purposes, beliefs, effort, management strategies and language learning strategies as the process. Precursor variables, in their model, are the unmodifiable learner variables of intelligence, aptitude, sex, age and prior learning. Even more distal to language achievement are the non-learner environmental (social, economic, cultural, etc.) and institutional variables including quality of instruction and school resources.

The Wen and Johnson model, like the Abraham and Vann model, does not recognize the roles of anxiety, motivation and attitude. In addition, there is no mention of cognitive styles, and aptitude plays a relatively minor role. Finally, both models can be criticized because their models show the relationships to be unidirectional and, therefore, do not explain how beliefs are developed and sustained.

Given the previous models of language learning and beliefs, as well as research exploring the relationship between beliefs and language learning, Figure 1 is a proposed model.

 

Abilities            Motivation

          Beliefs                          Attention Achievement

    Anxiety

Styles

                                        Attitudes

                                                          Strategies

Figure 1. Proposed Relationship Between Beliefs and Achievement

There are two salient characteristics of this model: 1) beliefs are central to the formation of affective factors and strategies; 2) the relationship between beliefs, strategies, affective, and achievement variables is bi-directional, whereas the relationship between beliefs and the unmodifiable variables is unidirectional. Following is a description of the foundations of this model.

The relationship between beliefs and strategies is clearer than that of the beliefs and affective variables. Wen and Johnson (1997) established that language learning strategies are directly related to the existence of beliefs. For example, students’ beliefs concerning the use of their mother tongue correlated strongly with their reported use of translation as a strategy. This, of course, is not surprising; that is, believing in a strategy to learn language is likely to be an important precursor to the initiation and sustained use of a strategy.

It should be pointed out, however, that the relationship between beliefs and actions is not necessarily a direct, causal one. Beliefs, by definition, are not constrained by external reality, whereas actions are. A student who does not believe in grammar-translation will still probably use grammar-translation oriented strategies if she is assessed by a grammar-translation test. Thus, while beliefs are likely to be the “mental triggers” to actions, they are limited by real-world constraints (Gaskell & Fraser, 1990).

The proposed relationship between beliefs, motivation, anxiety and attitudes is more tenuous. Several examples will explain the proposed relationship. One example is the belief that language study will lead to more career opportunities. This, in principle, is likely to influence a learner’s motivation to engage in language learning. An example of a belief that may have a negative relationship with motivation would be a belief that Japanese and males are not good language learners. This belief, if held by a Japanese man, would plausibly result in decreased motivation.

Beliefs might also affect a learner’s anxiety, especially if the learner’s belief system is significantly different from the instructional methods. For instance, Kern (1995) found that 31 percent of students in a French course agreed to the belief that “leaning a foreign language is mostly a matter of learning a lot of grammar rules,” yet no instructors endorsed the same belief. Students who believe in the utility of explicit grammar instruction will probably experience increased learning anxiety in a classroom led by an instructor who does not spend classroom time on explicit grammar instruction. Debilitative anxiety is a likely product of a classroom where the students’ and teachers’ beliefs about the nature of language learning are mismatched. Beliefs about the nature of communicative activities, public performance, grammar instruction, the need for negative evidence and the use of the students’ L1 in the classroom could lead to anxiety if there is a perceived mismatch.

The proposed relationship between attitudes and beliefs is, as of yet, not well defined. This is because belief and attitude scales often use very similar items. Baker’s (1992) study of attitudes toward bilingualism, for example, has items concerning the nature of language learning and the instrumental value of learning two languages—two topics covered in the Horwitz’s The Beliefs About Language Learning Inventory (BALI) (Horwitz, 1988) and the Sakui and Gaies (1999) questionnaire. Thus, the dividing line is not so clear.

There have been numerous studies describing the language learning beliefs of different cultural groups. Horwitz’ has done several studies looking at the beliefs of American undergraduate students studying commonly taught languages (Horwitz, 1987; Horwitz, 1988). Other research has examined Yemen students and teachers (Kuntz, 1997), Taiwanese (Yang, 1993; Banya & Cheng, 1997) and Japanese students (Sakui & Gaies, 1999).
 The Sakui and Gaies study is the only study to systematically look at language learners’ beliefs on a wide scale in Japan. Using a combination of original items and ones from previous questionnaires, mainly Horwitz’s BALI, they made a questionnaire in Japanese. Designed through a series of translations and back translations and repeated piloting, the questionnaire is the most rigorously constructed instrument available in Japanese.

The consistency or reliability of respondents’ choices was a great concern. In other words, if the instrument was not able to achieve consistent responses, one would have to question the basic constructs and the reliability of their beliefs measure. To a certain extent, the questionnaire did yield consistent responses, based on two administrations to the same sample. Respondents tended to be consistent in the direction of their choice—agree or disagree—but not the strength; that is, strongly agree or strongly disagree.

While the Sakui and Gaies study is an important step in creating a valid and reliable instrument, it did not examine the relationship between beliefs and other individual differences such as strategies, aptitude, motivation and anxiety. It is, however, an essential first step, one that should promote further research.

To date there has been very limited research on the most important issue concerning beliefs: Do they make a difference in language learning? Wen and Johnson (1997) found that beliefs do influence strategy choice. Indeed, the most direct link between beliefs and achievement is likely to be with strategies, but this would only be a beginning. Ellis (1994, p. 479) writes:
 

The study of learners’ opinions about language learning constitutes an important area of inquiry, as it is reasonable to assume that their ‘philosophy’ dictates their approach to learning and choice of specific learning strategies. Further research will need to find ways of ensuring that learners’ verbal report of their beliefs reflect their actual beliefs, and of investigating what effects different ‘philosophies’ have on learning outcomes.

 This study set out to answer three questions:
 

1. To what extent would the beliefs of a restricted sample of first year students at the end of the school year resemble that of Sakui and Gaies’ sample of mostly non-majors sampled throughout the school year?
2. Do beliefs vary systematically by gender, previous education, living and travel abroad experience?
3. Do beliefs vary systematically by English proficiency?

 

Method

Participants

Four intact classes of first year university students at Chuo and Sophia Universities participated in the study. One hundred and thirty-eight students completed part of the study, but there were data on all the measures available on 112 students. Appendix A gives the descriptive statistics for the participants.

Instruments

The main instrument used in this study is a questionnaire developed by Sakui and Gaies (1999). In addition a modified version of the Foreign Service Institute’s rating scale for oral proficiency (Hughes, 1994) was used to rate the oral proficiency of the participants (Appendix B). The adapted FSI rating scale has not been validated for the context in which it was used, and the speaking samples were rated only by the researcher. Consequently, the results from the scale should be interpreted with caution.
 

Procedure

The Sakui-Gaies belief questionnaire was administered to students during class time. All students were able to complete the instrument in less than 15 minutes.

As part of their regular classroom assessment, students in each class had to do a four to five-minute speaking test. In pairs or threes, students were instructed to speak, only in English, on any subject they wanted. Students were told before the assessment that they were going to be evaluated on paralinguistic measures (mainly eye contact and gestures), conversation flow (e.g., use of responsive remarks and questions) and naturalness, which was used to encourage students not to go through a pre-written memorized skit. The classroom evaluation criteria were not the ones used for this study, but were designed to encourage a natural and unrehearsed sample of speaking that was reflective of the students’ true proficiency. Despite the precautions to ensure natural, unrehearsed interaction, four groups were obviously going through a rehearsed script. All speaking tests were videotaped and analyzed using the modified FSI scale.
 

Results

Appendix C reports a comparison of the mean ratings for each item compared to the Sakui and Gaies results. The correlation between the means of this study and the Sakui and Gaies results was r = .92, (p<01), indicates that the mean ratings of the item are very similar. Thus, the overall differences in the item means indicate that the sample in this study is not substantively different from the Sakui and Gaies sample.

Furthermore, a factor analysis (with varimax rotation), indicates a strong similarity to the solution in the Sakui and Gaies study. (See Appendix D for complete results of factor analysis.) For example, Table 1 shows the first component (alpha = .52; 8.4% of variance).

---

---

These items all loaded on Factor 1 in the Sakui and Gaies study, which they labeled as “Contemporary (communicative) orientation to learning English.” Ten out of eleven of the items in the first factor are the same in both studies.

Given the descriptive results and the factor analysis, it is likely that there are few differences in the results of the Sakui and Gaies study and this one.

In terms of sex differences in beliefs, there are few significant differences. Item 10 [X2 (3, N = 112) = 10.04, p<0.5] and Item 33 [X2 (3, N = 112) = 8.0, p<.05] were statistically significant. Care should be taken, however, in interpreting these results because 45 chi-square tests were run, and so there is a great possibility of type I error (Brown, ms)

The other variables showed similar results; that is, very few significant differences. For instance, there was only one significant difference when comparing the questionnaire responses by travel abroad experience, Item 36 [X2 (3, N = 112) = 11.36, p<.05]. Items 12  [X2 (3, N = 112) = 9.91, p<,05] and Item 41  [X2 (3, N = 112) = 9.75, p<.05] were the only differences in beliefs by those who had and did not have non-compulsory schooling (i.e., cram schools, university exam preparation schools, and English conversation schools). A chi-square analysis could not be used for the Live Abroad data because the expected cell frequencies were less than five due to the small number of participants who had lived abroad (Norusis, 1998). Again, caution is necessary due to the multiple chi-square tests.

Thus, the answer to the second research question is that there appear to be no systematic differences in language learning beliefs by sex, travel abroad experience and non-compulsory education.

The third research question considered the differences in beliefs among different proficiency levels. The descriptive data for the proficiency measure are in Appendix E.

Since it has been established that participants do not vary much in their beliefs about language in relation to the other measures (sex, travel abroad experience and non-compulsory education experience), thus reducing the possibility of covariates clouding the analysis, Pearson-product moment correlation was used to look at the relationship between beliefs and proficiency. Table 2 shows the statistically significant correlations, and Appendix F gives the correlations of each item and the oral proficiency ratings.

---

Note: **  Correlation is significant at the 0.01 level (2-tailed).
     *  Correlation is significant at the 0.05 level (2-tailed).

---

These results indicate that more proficient speakers of English are more positive about their future English ability (Item 4), enjoy studying English more than less proficient students (Item 43), view culture learning as an important element of language learning (Item 5) and are more satisfied with the English education they have received (Item 44).

The more proficient participants, in contrast to their peers, believe that repetition and practice are not as important (Item 12) and are less agreeable to the propositions that translation is a prerequisite for learning (Item 33) and native speaker teachers should be able to communicate in Japanese (Item 39). In addition, they are more tolerant of mistakes (Item 35) and imperfect oral production (Item 6).

Conclusion

The first goal of this study was to replicate the Sakui and Gaies study, and to look for similarities between their large (N = 1296) sample and this researcher’s  teaching contexts. No systematic differences were found concerning basic agreement or disagreement with the propositions on the questionnaire, though several trends were evident. Furthermore, the basic factor structure that Sakui and Gaies found with their questionnaire was confirmed.

Participants’ responses did not vary according to sex, overseas travel experience and educational experience. While there were some differences in the results of the two studies, the use of multiple statistical tests and the relatively small sample size of the current study, make conclusions regarding these differences suspect. Indeed, it is recommended for a study to have at least 300 cases for a factor analysis (Tabachnick & Fidell, 1996).

The third research question examined the relationship between beliefs and proficiency. There were important differences on 10 of the items, with more proficient students agreeing with propositions that are likely to be a cause and consequence of their achievement. In contrast, the more proficient participants disagreed with propositions that are plausibly related to low achievement. This relationship, though weak, indicates that what one believes about language learning is related to achievement.

Causality, however, at this point is indeterminable. As is indicated in the proposed model in Figure 1, there is probably a bi-directional relationship between beliefs and other factors affecting achievement. For instance, success at language learning fosters beliefs about the nature of language learning that should lead to more achievement. In contrast, failure at language learning is likely to result in beliefs that hinder language achievement.

Future research will need to look at the relationship between beliefs, other factors and achievement as part of a dynamic system of relationships. Qualitative inquiry, including interviews and learner journals, as well as longitudinal designs would be able to answer some of the questions that cross-sectional questionnaire-based research is unable to answer.
 

References

• Abraham, R. &  Vann, R. (1990). Strategies of unsuccessful language learners. TESOL Quarterly, 24, 177-197.
• Baker, C. (1992). Attitudes and language. Avon, United Kingdom: Multilingual Matters.
• Banya, K. & Cheng, M. (1997). Beliefs about language learning—A study of beliefs of teachers’ and students’ cultural settings.  Paper presented at the Annual Meeting of the Teachers of English to Speakers of Other Languages (31st, Orlando, FL, March 11-15, 1997). ED 411 691.
• Brown, J.D. (1998). Using surveys in language programs. Unpublished manuscript.
• Ellis, R. (1994). The study of second language acquisition. New York: Oxford.
• Fraser, C. and Gaskell, G. (1990). The social psychology of widespread beliefs. Oxford: Clarendon Press.
• Gaies, S. (1998) Japanese language learners’ perceptions of methodological alternatives. Presentation at the Japan Association for Language Teaching 1998 Conference. (Omiya, Japan, November 20-23, 1998).
• Gardner, R., Tremblay, P. & Masgoret, A. (1997). Towards a full model of second language learning: An empirical investigation. The Modern Language Journal, 81, 344-262.
• Horwitz, E. (1987). Surveying student beliefs about language learning. In A. Wenden & J. Rubin (Eds.), Learning  strategies in language learning. Englewood Cliffs, NY: Prentice Hall.
• Horwitz, E. (1988). The beliefs about language learning of beginning university foreign language students. Modern Language Journal, 72, 283-294.
• Hughes, A. (1989). Testing for language teachers. Cambridge: Cambridge University Press.
• Kern, R. (1995), Students’ and teachers’ beliefs about language learning. Foreign Language Annals, 28, 71-92.
• Kuntz, P. (1996). Beliefs about language learning: The Horwitz model. ED 397 649.
• Norusis, M. SPSS 8.0 Guide to Data Analysis. Upper Saddle River, NJ: Prentice-Hall.
• Sakui, K. and Gaies, S. (1999). Japanese language learners’ perceptions of methodological alternatives. Manuscript submitted for publication.
• Tabachnick, B., & Fidell, L. (1996). Using multivariate statistics. New York: Harper Collins.
• Wen, Q., & Johnson, R.K. (1997) L2 learner variables and English achievement: A study of tertiary-level English majors in China. Applied Linguistics, 18, 27-48.
• Yang, N. (1993). Understanding Chinese students' language beliefs and learning strategy use. Paper presented at the Annual Meeting of the Teachers of English to Speakers of Other Languages (27th, Atlanta, GA, April 13-17, 1993). ED 371 589.

---

Appendix A. Descriptive Data on Participants

 

 

 

 

 

 

Appendix B. Modified FSI Oral Proficiency Rating Scale

Accent
1. Frequent gross errors and a very heavy accent make understanding difficult, require frequent repetition.
2. “Foreign accent” requires concentrated listening, and mispronunciations lead to occasional misunderstanding.
3. Marked “foreign accent” and occasional mispronunciations which do not interfere with understanding.
4. No conspicuous mispronunciation., but would not be taken for a native speaker.
 

Grammar
1. Constant errors showing control of very few major patterns and frequently preventing communication.
2. Frequent errors showing some major patterns uncontrolled and causing occasional irritation and misunderstanding.
3. Occasional errors showing imperfect control of some patterns but no weaknesses that cause misunderstanding.
4. Few errors, with no patterns of failure.
 

Interaction
1. Responsive to others’ comment, but does not initiate interaction.
2. Use of few major patterns, but unable to guide topics.
3. Able to ask questions and guide topic (s) to a limited extent.
4. Able to control and guide conversation.
 

Fluency
1. Speech is very slow and uneven except for short routine sentences.
2. Speech is frequently hesitant and jerky; sentences may be left uncompleted.
3. Speech is occasionally hesitant, with some unevenness caused by rephrasing or groping for words.
4. Speech is effortless and smooth, but perceptively non-native in speech and unevenness.
 

Comprehension
1. Unable to follow basic flow of topic. No verbal or physical indications of comprehension.
2. Several misunderstandings. Few verbal or physical indications of comprehension.
3. Few misunderstandings. Limited ability to show comprehension verbally or physically.
4. Able to follow the topic with no noticeable problems. Able to show comprehension verbally or physically.
 

***Note. Means based on four-point scale: 1=strongly disagree; 2=disagree; 3=agree; 4=strongly agree

 

Note. Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.
Only loadings of +/- .35 or greater are included in this solution.

 

Appendix E. Descriptive Data for Proficiency Measure