Attitude is one of the important affective factors mediating the SL/FL learning. Scholars, however, are divided as to what categories there are in it. This paper aims to develop and validate a survey of attitudes toward web-based autonomous college English learning (WBACEL). At the stage of pretesting, two teaching methodological experts were invited to have a first evaluation on the wording and format of the survey, and 20 items were suggested to delete. Then it is a small-scale pretesting; three more items without good enough discrimination were dropped out. Finally it is a large-scale pretesting; another ten items proved to be low correlated with the survey were deleted. At the stage of piloting, 283 valid questionnaires (out of total 318 students from nine regular classes) were collected and processed. From the results of item analysis (including item description, item discrimination analysis, and correlation analysis), and the test of construct validity, nine items were deleted. Factor analysis shows that the remaining 28 items can be divided into five attitudinal WBACEL factors: learning valence, resources and learning materials, learning plans and objectives, self-efficacy of learning, and evaluation of learning performance. The items of the whole questionnaire and those of each factor are consistent together with good indexes of Cronbach Alpha coefficient. The findings show a necessity to a reconceptualization of attitudes (in contrast to Gardner’s attitudinal construct in AMTB) in the web-based foreign language learning contexts.
The new College English Curriculum Requirements drawn up by the Ministry of Education [
The present paper designs a questionnaire to measure non-English major’s attitude of WBA English learning and tests its reliability and validity quantitatively. First it is the literature review, looking at the studies on the relation between students’ autonomous foreign language learning (FLL) attitudes toward web-based environment and those toward other FLL environments, and how researchers operationalized FLL autonomy (whether web-based or not) is also reviewed. Then the paper explains the procedure of data collection, including the sample selection, pretesting, pilot testing of the questionnaire and the way to analyze the data. Finally, the paper ends with discussing how the major findings are related to the results of the previous similar studies.
The environment of web-based (WB) FLL is believed to be beneficial to learners. Chapelle [
From the above description, it can be asserted that WB language learning environment can provide a natural context for learner autonomy, but such autonomy needs to be developed or fostered systematically (Zhu [
As is mentioned above, more and more researchers tend to extend the notion of FLL attitude to include more factors or scales. The same is true for those studying the autonomous FLL attitude. Little [
(a) To what extent the items in the web-based autonomous college English learning attitude survey (WBACELAS) are appropriate?
(b) What factors or scales of WBACEL attitude are there in the survey?
(c) To what extent the items (within factors and in total) in the survey are consistent?
The three questions are analyzed quantitatively. Question one is approached by item analysis (including item description and item discrimination analysis), and correlation analysis, question two by factor analysis and the third question by internal consistency reliability analysis (Cronbach Alpha coefficient).
Based on Zhu [
The population of respondents is the third-term non-English majors. The reason to choose them as the participants is that by more than one academic year of using CELP. They may develop a mature or near mature opinion on the effectiveness of the virtual learning platform.
The wording and format of initial version of the questionnaire was first evaluated by two professors majoring in English teaching methodology who have been giving the applied linguistic courses to postgraduates of English major, one having the course of teaching methodology, and the other learning strategies. The attitudinal variable of self-efficacy was suggested to delete because it was thought to overlap with the first variable of CELP valence and its items are added to the valence variable. Totally 20 items recommended to be repetitive with those in other variables were deleted and some items were reworded. Of those items deleted, 16 are in the attitudinal variable of CELP resources and learning materials, two in the attitudinal variable of participant role, and another two in learning evaluation variable. As to the items reworded, for example, an item that was originally worded as “With the study in the CELP, I make progress in my ability to autonomous English learning” was reworded as “With the study in the CELP, I find the progress in my competence of autonomous English learning” because the former was not about one’s attitude but fact of learning.
The 50-survey was then pretested on a small scale of ten students in the present author’s class, who are from the same population. Five of them are male, the other five female. Among these respondents there are top, mid- and low-level students. The purpose of the pretesting is to find out: (a) the wording problem, (b) the problematic items. To achieve the first goal, we had one sentence in black on the top of each page of the questionnaire: “Please underline where you find awkward or difficult to understand”. Some found “学伴” (peer students) in two items, for example, to be not natural. Therefore, it is reworded as “同班或者同宿舍同学” (classmates or roommates). As to the criteria of picking out problematic item, according to Dörnyei & Taguchi [
To further check the appropriateness of the survey, the 47-item questionnaire is conducted a second large scale pretesting. Some more negatively worded items (totally eight) are added behind their respective positively worded counterpart because they can pick out those uncooperative respondents in answering the questionnaire and these negative items do not go into the later data analysis. The administration was carried out in the first week of November in 2013 in the three regular classes of the present author6. The total number of students is 125, and questionnaires collected are 117 and 115 valid. To further find out problematic items quantitatively, a series of analyses is carried out: descriptive analysis, discrimination analysis and correlation analysis. First is the descriptive analysis. The item description shows no position response bias. The discrimination analysis through T test also does not tick out problematic items. To what extent a given item is proper can be also decided by looking at its coefficient (that is, how its mean score is correlated to the total mean score). Based on Qin [
After the pretesting, the 70-item questionnaire is reduced to 37 items (see
It is evident that the participant role attitudinal variable and the learning method attitudinal variable only have two items respectively. Anyway the pretesting helps effectively reduce the number of items so that they can be comfortably compiled within four A4 pages of a booklet of a questionnaire (Qin [
The WBACELAS was administered to 318 students, most of whom completed the survey during class time in the last week of November in 20138. These students are from nine classes. Four of them are on Campus A (centered on the arts), another four from Campus B (centered on the sciences), the last one class on Campus C. There is only one department (College of History and Tourism) on this campus. The detailed information is as follows in
Those questionnaires will be deleted if fulfilling any of the following criteria: (a) more than half of the information not provided, (b) all items answered in the way, or (c) 60 percent of negatively worded items being answered quite contradictorily to their positively worded ones. As a result 35 questionnaires are invalid. All the data are entered into the SPSS database. The analysis involves a descriptive analysis, discrimination analysis and
Attitudes towards CELP valence: 9 items |
---|
Attitudes towards WB resources and learning materials: 15 items |
Attitudes towards participant roles: 2 items |
Attitudes towards WBA learning objectives and plans: 5 items |
Attitudes towards WBA learning method: 2 items |
Attitudes towards evaluation of WBA learning performance: 4 items |
Classes Questionnaires | Campus A | Campus B | Campus C | Total | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Class 1 | Class 2 | Class 3 | Class 4 | Class 5 | Class 6 | Class 7 | Class 8 | Class 9 | ||
Collected | 36 | 34 | 31 | 28 | 40 | 40 | 42 | 28 | 39 | 318 |
Valid | 35 | 31 | 22 | 25 | 37 | 37 | 35 | 25 | 36 | 283 |
correlation analysis of each item, a construct validity test, and an internal consistency reliability test. The descriptive analysis, discrimination analysis and correlation analysis of each item allow us to detect if there are any problematic items before grouping them into different categories. The test for construct validity allows us to reduce the data by means of an exploratory factor analysis through the use of the principal components extraction and Varimax rotation methods. The test for internal reliability (Cronbach’s Alpha) allows us to calculate the Cronbach Alpha levels for each factor and the whole items.
The descriptive result of the items (only positively worded ones) is shown in the following
The following
From T-test, it seems that all (positively worded item) are not problematic. Correlation analysis reveals, however, that item 21, 29, 30, 39 and 40 are not proper because their Spearman coefficients are less than 0.3. Therefore, these five items are dropped in the following validity analysis.
Item | Scale | Valid | Missing | Item | Scale | Valid | Missing | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
(1)* | (2)* | (3)* | (4)* | (5)* | (1) | (2) | (3) | (4) | (5) | ||||||
1 | 19.4 | 29 | 39.6 | 10.2 | 1.4 | 282 | 1/0.4** | 22 | 8.1 | 14.8 | 36.6 | 30.6 | 9.5 | 282 | 1/0.4 |
2 | 19.1 | 33.2 | 24.4 | 16.3 | 2.5 | 270 | 13/4.6 | 23 | 6.3 | 20.8 | 37 | 29.6 | 6 | 282 | 1/0.4 |
3 | 9.9 | 30.3 | 37 | 18.3 | 1.4 | 274 | 9/3.2 | 25 | 9.9 | 26.4 | 43 | 16.9 | 2.5 | 279 | 4/1.4 |
4 | 25.7 | 37.7 | 25 | 9.5 | 1.8 | 282 | 1/0.4 | 26 | 6.3 | 25.4 | 37.7 | 26.8 | 3.5 | 282 | 1/0.4 |
5 | 17.3 | 35.6 | 29.9 | 14.1 | 2.8 | 282 | 1/0.4 | 28 | 7 | 16.2 | 35.9 | 28.5 | 12 | 282 | 1/0.4 |
6 | 20.4 | 36.6 | 29.6 | 10.9 | 2.1 | 282 | 1/0.4 | 29 | 13.4 | 27.5 | 30.3 | 21.1 | 7.4 | 282 | 1/0.4 |
7 | 18.3 | 34.5 | 32.7 | 11.3 | 2.5 | 283 | 0 | 30 | 6.7 | 21.5 | 33.8 | 26.4 | 11.6 | 283 | 0 |
8 | 20.1 | 32.7 | 32.4 | 11.3 | 1.8 | 281 | 2/0.7 | 32 | 18.7 | 36.3 | 29.9 | 11.3 | 3.5 | 282 | 1/0.4 |
10 | 17.3 | 23.2 | 37.7 | 16.5 | 4.9 | 282 | 1/0.4 | 34 | 11.6 | 30.3 | 45.4 | 10.2 | 2.5 | 283 | 0 |
12 | 11.3 | 22.2 | 25.7 | 30.3 | 10.2 | 282 | 1/0.4 | 35 | 12 | 34.2 | 39.4 | 11.3 | 2.8 | 282 | 1/0.4 |
13 | 18.3 | 34.9 | 24.6 | 16.2 | 3.5 | 276 | 7/2.5 | 37 | 9.2 | 33.5 | 37 | 15.5 | 4.6 | 282 | 1/0.4 |
14 | 15.1 | 20.4 | 32.7 | 22.5 | 9.2 | 283 | 0 | 38 | 6.3 | 14.1 | 32 | 38 | 9.2 | 282 | 1/0.4 |
15 | 7 | 16.9 | 30.6 | 34.2 | 10.9 | 282 | 1/0.4 | 39 | 5.3 | 9.5 | 24.6 | 39.1 | 20.1 | 279 | 4/1.4 |
16 | 16.2 | 29.6 | 30.6 | 17.3 | 6 | 282 | 1/0.4 | 40 | 5.6 | 12.7 | 22.9 | 36.6 | 20.8 | 279 | 4/1.4 |
17 | 8.5 | 14.4 | 23.6 | 34.5 | 19 | 283 | 0 | 42 | 9.9 | 19.7 | 31 | 30.3 | 8.1 | 280 | 3/1.1 |
18 | 15.1 | 21.8 | 28.5 | 27.5 | 6.7 | 282 | 1/0.4 | 43 | 14.1 | 22.9 | 29.6 | 27.8 | 4.6 | 280 | 3/1.1 |
19 | 17.6 | 24.6 | 31.3 | 19.7 | 6.7 | 283 | 0 | 44 | 8.1 | 16.9 | 34.5 | 32.4 | 7 | 280 | 3/1.1 |
20 | 12 | 18.3 | 31.3 | 27.1 | 10.9 | 282 | 1/0.4 | 45 | 7 | 20.8 | 36.3 | 28.2 | 6.7 | 280 | 3/1.1 |
21 | 4.6 | 8.1 | 16.9 | 36.6 | 33.8 | 283 | 0 |
*1 = The statement is never or almost never true of me, 2 = The statement is not true of me, 3 = The statement is sometimes true of me, 4 = The statement is true of me, 5 = The statement is totally or almost totally true of me; **The number behind “/” is percentage.
Item | T | R | Item | T | R | Item | T | R |
---|---|---|---|---|---|---|---|---|
1 | 25.748** | 0.526** | 16 | 30.158** | 0.537** | 32 | 24.97** | 0.422** |
2 | 29.155** | 0.649** | 17 | 30.83** | 0.471** | 34 | 21.111** | 0.435** |
3 | 26.299** | 0.526** | 18 | 37.188** | 0.484** | 35 | 21.875** | 0.403** |
4 | 32.552** | 0.536** | 19 | 37.23** | 0.595** | 37 | 24.758** | 0.427** |
5 | 26.024** | 0.604** | 20 | 20.522** | 0.487** | 38 | 24.513** | 0.334** |
6 | 26.46** | 0.59** | 21 | 30.467** | 0.242** | 39 | 24.577** | 0.247** |
7 | 25** | 0.574** | 22 | 27.209** | 0.544** | 40 | 27.986** | 0.198** |
8 | 25.861** | 0.554** | 23 | 35.957** | 0.441** | 42 | 34.955** | 0.414** |
10 | 29.334** | 0.458** | 25 | 25.118** | 0.391** | 43 | 37.53** | 0.411** |
12 | 35.31** | 0.439** | 26 | 37.899** | 0.401** | 44 | 29.984** | 0.39** |
13 | 28.468** | 0.512** | 28 | 27.978** | 0.398** | 45 | 35.534** | 0.368** |
14 | 36.818** | 0.499** | 29 | 36.017** | 0.088 | |||
15 | 29.29** | 0.544** | 30 | 35.536** | 0.182** |
**Difference or correlation is significant at 0.01 level (2-tailed).
The remaining 32 items are then subjected to principal components factor analysis to examine the internal structure. The factorability test of the data indicates a good factor analysis because the KMO value is 0.864, exceeding the recommended value of 0.6 and the Barlett’s Test of Sphericity reaches statistical significance (p = 0.000). From exploratory factor analysis and the interpretability of the resulting factors, a five-factor solution is most appropriate. These five factors accounts for 51.226 percent of the common variance10. The items with greater than 0.3 loadings determine the interpretation and labeling of the factors. The first factor receives appreciable loadings from 9 items and accounts for 15.017 percent of the common variance. This factor, including items 1, 2, 3, 4, 5, 6, 7, 8, and 10, falls exactly into the attitudinal variable of CELP valence in the initial WBACWLAS. Therefore, the factor can best correspond to the Attitudes toward WBA Learning Valence.
The second factor receives appreciable loadings from 10 items and accounts for 12.962 percent of the common variance. The items included in this factor are 12, 13, 14, 15, 16, 17, 18, 19, 22, and 28. They are in the attitudinal variable of CELP resources and learning materials in the initial WBACWLAS. This factor, therefore, can well correspond to Attitudes towards WB Resources and Learning Materials.
Factor three has appreciable loadings from 5 items and accounts for 8.594 percent of the common variance. This factor, including items 32, 34, 35, 37, and 38, falls exactly into the attitudinal variable of learning objectives and plans in the initial WBACWLAS. Therefore, this factor seems to well reflect a dimension of Attitudes toward WBA Learning Objectives and Plans.
Factor four receives appreciable loadings from 3 items and accounts for 7.372 percent of the common variance. The items included in the factor are 23, 25, and 26. It is separated from the attitudinal variable of WBA learning materials and resources in the initial WBACWLAS. The three items are respectively about the attitude toward the extent which CELP learning materials can help develop the English skills of listening, speaking and reading. This factor, therefore, can correspond to Attitudes toward WBA Learning Self-efficacy.
Factor five has appreciable loadings from 4 items and accounted for 7.282 percent of the common variance. This factor, including items 42, 43, 44, and 45, falls exactly into the attitudinal variable of WBA learning evaluation in the initial WBACWLAS. Therefore, this factor can best reflect a dimension of Attitudes toward the Evaluation of WBA Learning.
Moreover, based on Cid, Grãnena & Tragant [
To examine the cohesion of the items, a test of reliability is carried out with the remaining 28 attitudinal items. As is shown in the Appendix, the corresponding Cronbach Alpha level of each of the five scales, except for the scale of attitude toward WBA learning evaluation, are above 0.7, and the Cronbach Alpha level of all 28 item is 0.8963. The results show that the items have good internal consistency.
Summing up, after the item analyses (including item description, discrimination, and correlation), validity and reliability tests conducted in this section, a total of nine items have been deleted and consequently in the final version of the WBACELAS, and there are now 28 items (see Appendix for details). These correspond to five factors with the following indexes of reliability. Factor I, labeled “WBA Learning Valence” attitude, refers to students’ acknowledgement of the importance of WB autonomous English learning and the significance of such learning in the cultivation of one’s study habit, ability to C-E or E-C translation and so on. This factor includes six items and has a Cronbach Alpha level of 0.8242. Factor II, labeled “WB Resources and Learning Materials” attitude, can be defined as the satisfaction of CELP software facilities and an interest in the materials of various sub-platform (such as those in the English Course Platform, Tourism English Bank, Yaxinda English Bank and so on). This factor includes 10 items and has a Cronbach Alpha level of 0.86. Factor III, labeled “WBA Learning Objectives and Plans” attitude, refers to an evaluative opinion on the time planned for the WBA learning, the appropriateness of such plans, the objectives and its effectiveness, and the monitoring of learning procedure. This factor includes five items and has a Cronbach Alpha index of 0.7604. Factor IV, labeled “WBA Learning Self-efficacy” attitude, is defined to be how CELP achieves one’s goal in the cultivation of listening, speaking and reading skills. This factor includes three items and has a Cronbach alpha index of 0.7435. Finally, Factor V, labeled “Evaluation of WBA Learning” attitude, refers to students’ opinion on how the WBA learning outcomes should be evaluated. This factor includes four items and has a Cronbach Alpha level of 0.6639.
After the pretesting and the piloting, some change takes place for the questionnaire: (a) it is shortened to gain internal consistency and good discrimination of items, (b) the initial variables are restructured and redefined to gain validity. In this section we make a comparison between the initial variables with the final categories resulting from the validation, and how final categories are related with what previous studies found is also discussed.
All the nine items in the attitudinal variable “WBA Learning Valence” of the initial WBACWLAS have proved to load on one same construct (Factor I). However, three deleted items making reference to the assisting role of WBA learning to listening, speaking, and reading are repetitive with those in a new construct resulting from the factor analysis. Most items in the variable “attitude toward WB resources and learning materials” of the initial WBACWLAS also prove to load on one same construct (Factor II), but its three items are separated by factor analysis to form a new construct. All the remaining items in the attitudinal variable “WBA Learning plans and objectives” of the initial WBACWLAS have also proved to load on one construct (Factor III). Attitudinal factor IV “WBA Learning Self-efficacy” is a new construct, which was suggested to delete at the stage of expert assess of the questionnaire. The four remaining items in the attitudinal variable “Evaluation of WBA learning performance” of the initial WBACWLAS have proved to load on one same construct (Factor V). The attitudinal variables of “Participant Role” and “WBA learning method” in the initial WBACWLAS have dropped out by factor analysis. They are deleted mainly because a large proportion of their items are dropped out in pretesting by expert evaluation, item analysis, and correlation analysis (see
Compared these factors with those published in previous research, a number of coincidences and departures can be found. Factor I “Attitude toward WBA Learning Valence” partly coincides with Gardner’s [
Factor II labeled “Attitudes towards WB resources and Learning Materials” and factor V, labeled “Attitude toward Evaluation of WBA Learning” can be partially equated with Zhu’s [
The core concept of factor IV “Attitude toward WBA Learning Self-efficacy” can also be found in studies such as Qin & Wen [
This study is to elaborate a structured questionnaire to measure attitudes of WB autonomous English learning for the college non-English majors. After the stages of pretesting and piloting, the initial survey, which included 70 items and seven variables, has undergone several changes. It now includes 28 items which cluster into five factors (obtained by factor analysis) with good internal consistency. However, the questionnaire should never be taken granted as a valid instrument without validation in the subsequent studies. Firstly it is because the virtual college English learning center of most universities is school-based, and the present instrument may be subjected to some changes of items when used for similar research purpose. Secondly the validity of the factors will need to be confirmed in the subsequent studies where larger samples of college learners use WB mode of autonomous learning of English as a foreign language. In any case, the data in this study seem to be a necessity to a reconceptualization of attitudes (in contrast to Gardner’s [
This study was funded by the 2011 Projects of Educational Science in the Twelfth Five-Year Plan sponsored by the Educational Department of Guangxi Zhuang Autonomous Region, the serial number of project being 2011ZY002.
Huaikui Li, (2015) Developing and Validating Survey of Attitudes toward Web-Based Autonomous College English Learning. Open Access Library Journal,02,1-12. doi: 10.4236/oalib.1101543
Results of factor analysis and internal consistency analysis of WBACELAS items: Rotated Eigenvalues and Cronbach Alpha Coefficients (a)
a of overall items = 0.8963 (28 items). Key: Items 3, 4, 5 and 20 are deleted from the final version of WBACELAS.