EXAMINING MODEL OF ENGLISH FOREIGN LANGUAGE PROFICIENCY USING PLSPATH INWARD MODE.pdf PDF

Preview

Summary

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 2 Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. Interna...

Description

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 2

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 3

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 4

EXAMINING MODEL OF ENGLISH FOREIGN LANGUAGE PROFICIENCY USING PLSPATH: INWARD MODE Ratna Rintaningrum [email protected] Institut Teknologi Sepuluh Nopember (ITS), Surabaya Jl. Arief Rahman Hakim, Keputih, Sukolilo, Surabaya Abstract: The advancement of skills in the learning of English as a Foreign Language as distinct from learning English as a Second Language, presents some serious challenges for teachers in Non-English speaking countries, in particular, in Indonesia. This is particularly the case when seeking to measure change in learning of English over time. A major concern for researchers is the lack of information in the literature about the learning of English as a Foreign Language in Asian countries at the tertiary level, when compared with the volume of literature available on the learning of English as a Second Language in Western secondary schools. The study presented here goes some way to redressing this gap. The study makes use of secondary data gathered from an Indonesian University concerning the measures of three English skills, namely those of listening, writing, and reading that are tested on the three different occasions. The primary purpose of the study is to examine whether these three skills can be measured using the least squares strategy of statistical analyses employed by the PLSPATH computer program. Several models are examined with particular emphasis given to a model developed in the inward mode. By so doing it is hoped that proficiency in the teaching and learning of English as a Foreign Language in an Indonesian tertiary setting can be better assessed. Keywords: English as a foreign language, proficiency, secondary data, change over time, Partial Least Squared Path (PLSPATH)

INTRODUCTION The emergence of English as a World language is now indisputable. Crystal (2000) and Nunan (2001) as well as British Council (2013) argued that the spread of English provided unlimited access to the modern world of science, information and communications technology (ICT),

money,

power,

international

communication, and

intercultural

understanding as well as entertainment and many more fields. English has been said to have official status in 60 countries as the second language and has a prime place in 20 more countries as the major foreign language (Yang, 2001). It is widely recognised that English is the native language of five countries: the United States, the United Kingdom, Australia, New Zealand, and Canada.

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 253

This paper is concerned with measuring only three skills of English Foreign Language proficiency, namely, listening, writing, and reading that are tested on three different occasions to provide test scores, namely, the PRETEST, DIAGNOSTIC, and ELPT scores. PRETEST, DIAGNOSTIC, and ELPT are acronym used to define the latent construct used in the model. This paper employs variables and examines relationships between variables that relate to the components of English Foreign Language Proficiency Tests, namely, Listening Comprehension, Structure and Written Expression and Reading Comprehension.

Method Population The available population for this study is (a) all undergraduate students, (b) who enrol in the advanced English (English 2 course) and have undertaken the English II final test (c) at the University involved, (d) during the period 2007-2009. There are about 1975 students who form the target population participating in this study.

Modelling with PLSPATH Partial least squares path (PLSPATH) analysis is a general technique for estimating path models and is extremely useful in situations with massive amounts of data, but relative scarcity of theoretical knowledge (Sellin, 1995). The PLS strategy requires the development of an appropriate causal model and the testing of that model (Sellin, 1990; 1995; Falk and Miller, 1992). The causal model involves latent constructs (indirectly observed) and manifest variables (directly observed), which specify the inner and outer model relationships, respectively. Modelling with PLSPATH explores the relationships between both latent and manifest variables (Noonan & Wold, 1988; Rigdon, 2012). In particular, Rigdon (2012) argued that PLSPATH is able to examine models with latent variables that are constructed in three different modes, namely the outward, the unity and the inward modes. The main research question in this section is: 1. Do the three skills form proficiency in the use of the English language?

Outer Model PLSPATH Result: Inward Mode Pretest (PRETEST)

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 254

In this second model, the LV PRETEST is formed by three manifest variates, namely Listen1, Write1 and Read1. The outer model results in Table 1.1 show that there are sizeable differences in the values of the standardized regression weights among the MVs of Listen1, Write1 and Read1. Table 1.1 Outer Model Results for Model of English Language Proficiency: Inward Mode Outer Model PLS Inward Mode Weights (β) and Loadings (λ) a Pretest Listen1 PRETEST Write1 Read1 Diagnostic Listen2 NOSTIC Write2 Read2 Proficiency Listen3 ELPT Write3 Read3

Time 1 Weight 0.47 0.33 0.40

Time 2 Loading 0.84 0.80 0.84

Weight

0.55 0.40 0.24

Time 3 Loading

Weight

Loading

0.90 0.83 0.71 0.66 0.00 0.50

0.90 0.64 0.82

a

These coefficients are standardized regression coefficients with small jackknife standard error values that indicate the effects of multicollinearity and problems of rounding involved in estimation.

The manifest variate of Listen1 appears to contribute strongly to the formation of the PRETEST construct with a weight of 0.47 followed by Read1 with a weight of 0.40. Write1 has the lowest contribution to the formation of PRETEST with a weight of 0.33. Consequently, out of the three MVs that form the PRETEST construct, Listen1 can be considered to have the strongest contribution to the PRETEST construct in the inward mode. It must be noted that the tolerances do not exceed 0.50, and there are no problems arising from suppressor effects. The figure for this inward mode model of English Language Proficiency are presented in Figure 1.1.

Figure 1.1 Model of the English Language Proficiency: Inward Mode

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 255

Diagnostic Test (NOSTIC) The outer model results in Table 1.1 show that there are sizeable differences in the values of the weights among the manifest variates of Listen2, Write2, and Read2 between the two occasions of Time 1 and Time 2. It can be seen from Table 13.8 that Read2 has a weight of roughly a half of the weights of Listen2 and Write2, since Read2 has the smallest weight with a value of 0.24, while Listen2 has the greatest weight with the value of 0.55. Write2 also has a sizeable increase in weight with β= 0.40. Not only does Listen2 have the highest value of weight at Time 2, it also records the largest increase in weight between Time 1 and Time 2. Moreover, Read2, not only has the lowest weight at Time 2, it also has a substantial decline in its contribution to the structure of English Foreign Language Proficiency between Time 1 and Time 2. Similar to Time 1, Listen2 is the strongest manifest variate with the greatest increase in its contribution to the NOSTIC construct in the inward mode. This change appears to be a consequence of the opportunities provided for the development of the skills of listening through practice in a language laboratory and other factors that help increase the scores of listening comprehension of students between Time 1 and Time 2, but this is associated with a decline in the contribution of Reading to the structure of English Language Proficiency over the same time period of Course 1. Proficiency (ELPT) Proficiency (ELPT) is also hypothesized to be formed by three manifest variates, namely Listen3, Write3, and Read3. Table 13.8 shows that with respect to the other Times (Time 1 and Time 2), at Time 3 Listen3 and Read3 have the sole contributions to the structure of ELPT as indicated by the weights of 0.66 (0.01) and 0.50 (0.01) respectively. Surprisingly, Write3 no longer contributes, since it has a non-significant or zero weight and with problems associated with multicollinearity and a suppressor effect. Thus, Write3 is dropped from the ELPT construct. Among the manifest variates that form the ELPT construct, Listen3 can be considered to be the strongest contributor to the ELPT construct with an increase in its weight between Time 2 and Time 3 during Course 2. Moreover, Read3 increases greatly in its contribution to the structure of ELPT between Time 2 and Time 3, and this appears to be a consequence of greater emphasis placed on Reading during this period, but this seems to be at the expense of the contribution of writing skills that may arise from the teaching in Course 2.

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 256

Inner Model PLSPATH Result: Inward Mode Diagnostic Test (NOSTIC) In the path model in Figures 1.2 NOSTIC is hypothesized to be influenced by the LV PRETEST. The results of the analysis indicate that the variable PRETEST has significant effects on the diagnostic test (NOSTIC) with a beta coefficient of 0.61 (0.00) and correlation coefficient of 0.61 in the outward mode model, while in the inward mode model PRETEST has a slightly higher value of the beta coefficient of 0.62 (0.03) and correlation coefficient of 0.62. The inner model results of analysis in the inward mode produces a marginally larger path effect than in the outward mode, as a consequence of the fact that the weights are estimated to maximize the variances explained. The results in Table 1.2 also indicate that the R-square index for NOSTIC is 0.38 in the inward mode which indicates that the variances explained for this variable in the inward modes is 38 per cent respectively, which involve residuals paths presented in Figure 1.1 for NOSTIC of 0.79 in both cases. Table 1.2 Inner Model Results for the Models of English Language Proficiency Inner Model Results

Inward Mode

PLSPATH

Weight (β)

JknStd (se)

Corr (r)

Dependent Variable Independent Variable

NOSTIC PRETEST

0.62

0.03

0.62

Dependent Variable Independent Variable Independent Variable

ELPT PRETEST NOSTIC

0.19 0.35

0.03 0.03

0.41 0.47

Pred LV

R²

Q²

1 2

0.38 0.24

0.38 0.24

Variance Dependent Variable Dependent Variable

NOSTIC ELPT

Proficiency Test (ELPT) In the proposed model presented in Figure 1.2 ELPT is hypothesized to be influenced by two LVs, namely, PRETEST and NOSTIC. ELPT involves IRT-scaled scores that are calculated as an average. The results of the inner model analysis indicate that the variable NOSTIC has a greater beta coefficient than the variable PRETEST, this probably arises from its more proximal nature with respect to time of operation. The LV NOSTIC has a very significant effect on ELPT in the inward mode model NOSTIC has a beta coefficient of 0.35 (0.03) and a correlation coefficient of 0.47. The value of the beta coefficient for the LV Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 257

PRETEST is of 0.19 (0.03) and with correlation coefficient of 0.41 in the inward mode model. The results in Table 1.2 also indicate that the R-square index for the ELPT is 0.24 in the inward mode model which indicates that the variance explained for this variable in the inward mode model is 24 per cent respectively. This involves residual paths presented in Figures 1.2 for ELPT of 0.87.

CONCLUSION AND SUGGESTION In conclusion, the variables that are hypothesized to influence the final English Foreign Language Proficiency Test (ELPT) have significant effects on their dependent variables. For example, PRETEST has a very significant effect on NOSTIC. This indicates that students who have good scores in the PRETEST perform better in the diagnostic test (NOSTIC). It is important to note that Course 1 is conducted between PRETEST (Time 1) and NOSTIC (Time 2). The availability of Course 11 enables students to obtain better scores in NOSTIC. It is also important to note that Time 2 (NOSTIC) is at the beginning of Course 2 2. Thus the length of time between T1 and T2 may vary greatly between students, and this possibly influences students’ scores obtained at Time 2 (NOSTIC). It is because some students take NOSTIC straightaway after their completion of Course 1, but some students take NOSTIC at later years closer to their graduation. Students who take NOSTIC straightaway after completing Course 1 obtain more benefits than students who take NOSTIC at later years. The reason is probably that students still remember their knowledge about the English language taught in Course 1, and this enables them to have better scores in English. Moreover, there are two predictors of ELPT (T3), namely PRETEST (T1) and NOSTIC (T2), and it is recorded in Table 1.2 that NOSTIC (β=0.35) has a stronger direct effect than PRETEST (β=0.19) on ELPT. This indicates that the availability of Course 2 that is conducted between NOSTIC and ELPT enables students to gain better scores in English. This implies that students who take NOSTIC are more likely to have better scores at their final test (ELPT). Since NOSTIC is at the beginning of Course 2, and the availability of

1 2

English 1c English 2c

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 258

Course 2 enables students to gain higher scores in English, consequently, Course 2 is important and its availability should be continued, and not omitted. However, the results of inner model in Table 1.2 indicate that the beta coefficient of the direct effects from NOSTIC to ELPT (β=0.34) is smaller than the effect from PRETEST to NOSTIC (β=0.61). This suggests that English 2c (Course 2) that is conducted between Time 2 (NOSTIC) and Time 3 (ELPT) has greater impact on students with a lower level of English Language Proficiency. This also indicates that if these students just take PRETEST then take ELPT, without taking NOSTIC, they do not gain scores that are as high as students who take NOSTIC. The positive relationship between NOSTIC and ELPT also implies that the higher the scores students obtain on NOSTIC, the better the scores students obtain on ELPT. Therefore, students who have already had good performance in English are more likely to have higher scores in their final test if they take advantage by enrolling in Course 2. Some students who attain an appropriate standard of performance of graduation did not take English 2c and their scores at T2 was reassigned as their score for T3. PRETEST also has a direct effect on ELPT (0.19) in the inward mode operating through NOSTIC produces the total effect of 0.41. This total effect of PRETEST (0.41) on ELPT is necessarily larger than the direct effect of PRETEST on ELPT. This result of the analysis also raises the possibility that there are students who do not take Course 2 since these students probably had good performance in English. This argument can be explained through the effects of Time 2 (NOSTIC) on Time 3 (ELPT) that are smaller than the effects between Time 1 (PRETEST) and Time 2 (NOSTIC). These results suggest that Course 2 assists students with a lower level of English Language Proficiency to gain better scores. The total effect of PRETEST (T1) on ELPT (T3) implies that students with a higher level of English Language Proficiency possibly do not take Course 2, and this is allowed by the University, consequently, the effect of NOSTIC on ELPT is smaller than the effect of PRETEST on NOSTIC. Therefore, these results of inner model analyses may be interpreted in several ways, and they are important in order that policies can be better shaped for the future benefits for the University under survey, for other universities in Indonesia, and surrounding countries.

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 259

Since PLSPATH provides three different strategies, namely unity, inward, and outward modes in its operation, it is important to do further analyses in the unity and the outward mode model. REFERENCES Crystal, D. (2000). Emerging Englishes. English Teaching Professional, 3-6. British Council (2013). The English Effect. Nunan, D. (2001). English as a Global Language. TESOL Quarterly, 35, 605-606. Yang, R. (2001). An obstacle or a useful tool? The role of the English language in internationalizing Chinese universities. Journal of Studies in International Education, 5(4), 342-358. Rigdon, E. E. (2012). Rethinking Partial Least Squares Path Modeling: In Praise of Simple Methods. Long Range Planning, 45, 341-358. http://dx.doi.org/10.1016/j.lrp.2012.09.010 Wold, H. (1982). Soft Modelling: The basic design and some extension. In K. G. Joreskog & H. Wold (Eds.), System under indirect observation. Part III. Amsterdam, The Netherland: North Holland Press. Sellin, N. (1990). PLSPATH Version 3.01: Program Manual. Hamburg, Germany. Sellin, N. (1995). Partial Least Square Modelling in Research on Educational Achievement. In W. Bos, Lehmann, R.H. (eds) (Ed.), Reflection on Educational Achievement: Paper in Honour of T. Neville Postlethwaite (pp. pp. 256-267). Waxmann, New York. Falk, F. R. (1987). A Primer of Soft Modelling. Institute of Human Development, University of California, Berkeley. Keeves, J. P., & Sellin, N. (1994). Path Analysis with latent variables. In T. Husén & T. N. Postlethwaite (Eds.), The International Encyclopedia of Education (2 ed., pp. 43524359). Oxford: Pergamon Press.

Ratna Rintaningrum (2015). Examining Model of English Foreign Language Proficiency Using PLSPath: Inward Mode. International Conference Proceeding, Udayana University: Bali. Page 260...