Differences in Attitude and Behaviour towards

Learning English Monolingually and Bilingually:

A Rasch Analysis with Banda Aceh Data

Khairiah Syahabuddin

Department of English Language Education, Faculty of Education and Teacher Training,

Ar-Raniry State Islamic University (UIN Ar-Raniry), Banda Aceh – Indonesia And

Researcher of Centre for Area Studies (CfAS) Ar-Raniry State Islamic University (UIN Ar-Raniry), Banda Aceh –

Indonesia

Keywords: Behaviour in EFL learning, Gender, Bilingual and Monolingual Learning, Rasch Analysis.

Abstract: Many educational variables like motivation, self-concept, self-regulation, learning and so on, have been

measured using attitudes and their corresponding behaviours with the same items on a linear scale in the

past two decades using Rasch analysis. Supported by the Rasch measures, the theory showed that each

attitude item was easier than its corresponding behaviour item. The present study was participated by 780

male and female first-year middle school students (12-13 years old) in 2011, consisting of 394 students

taught in bilingual schools and 386 students taught in monolingual schools, which were selected from a

number of schools with bilingual and monolingual teaching programs in Aceh. The result of the study

showed that attitudes were not always easier than their corresponding behaviours, apparently because

students taught monolingually have different attitudes to learning English than bilingually taught students,

although both groups of students have similar behaviours towards learning English, as determined by a

Rasch measure. This is an interesting finding not previously reported and suggests that Rasch measures of

student attitudes by monolingual and bilingual teaching should be further investigated.

1 INTRODUCTION

Aceh province in Indonesia experienced a huge

Tsunami in 2004 that involved a massive loss of life

(over 220,000 were killed) and almost total

destruction of infrastructure and housing. A lot of

help and support were offered by Indonesia and the

international community, including both financial

and material support. One aspect of international

supports was the introduction of bilingual education

to Aceh province which was part of the newly-

implemented curriculum 2004 launched by the

Indonesian Ministry of Education in Jakarta. The

establishment of formerly standard schools into

schools with bilingual programs ran well. Some

internationally funded organisations from Turkey,

for example, successfully established new private

bilingual dormitory schools, with Turkish and

English as the main languages used. Public schools

(Sekolah Negeri) with bilingual programs started to

gain popularity as well. Both private bilingual

dormitory schools and public schools with bilingual

programs became favourites with the local

population (anecdotal evidence from the author).

Most of the Aceh parents tried to enroll their

students in the bilingual schools and felt unhappy if

their children were rejected.

However, there was no a scholarly evidence that

students who studied in schools with bilingual

programs in Aceh performed better than their

counterparts studying at the standard monolingual

schools. There was anecdotal folk-rumour that

students studying at public schools with bilingual

programs out-performed students studying at public

schools with monolingual programs, yet there was

no published evidence for this conclusion.

This study was part of larger study which

investigated English language achievement of first-

year Middle School students in Banda Aceh,

Indonesia, regarding their ability in English Reading

Comprehension, English Writing, and their

Behaviour towards Learning English regarding their

experience in learning English as a foreign language,

160

Syahabuddin, K.

Differences in Attitude and Behaviour towards Learning English Monolingually and Bilingually: A Rasch Analysis with Banda Aceh Data.

DOI: 10.5220/0008215100002284

In Proceedings of the 1st Bandung English Language Teaching International Conference (BELTIC 2018) - Developing ELT in the 21st Century, pages 160-170

ISBN: 978-989-758-416-9

in context of difference in type of schools and

Gender. The aim of this study was to investigate

students’ attitude and behaviour towards learning

English bilingually and monolingually in Banda

Aceh. This paper reports on the measurement of

students’ English behaviour using Rasch analysis

with the RUMM2030 computer program (Andrich et

al. 2010).

English used in bilingual classes is around 50%

with another 50% for Bahasa Indonesia. English

used in monolingual classes is around 20% with the

rest 80% for Bahasa Indonesia. For the purpose of

this study, a bilingual school program is defined as

the teaching of English by using a combination of

English and Bahasa Indonesia with equal percentage

of time, and the monolingual school program is

defined as the teaching of English by using a

majority of Bahasa Indonesia and a small amount of

English by time. Except for the medium of language

instruction, bilingual and monolingual groups

maintained all other conditions and situations the

same. Both groups shared the same

curriculum/syllabi, the same length of learning time,

the same textbooks, the same teacher’s qualification,

the same kind of materials for homework, and the

same English activities at school.

Teachers in both types of schools were

counselled and monitored during the experimental

study in order to ensure that there were no changes

in conditions and situations and that all schools were

working the same way throughout the two months of

study. The two months were chosen due to two

considerations. First is the study or English content

issue. This is related to students’ exposure to outside

classroom or outside school activities that could

confound the experiment and call the conclusions

into question. Should any types of students learn

extra English content which differentiated them

from the other group, the conclusions of the study

could be called into question. So, outside school

English learning (such as homework) was closely

monitored. Second is teacher willingness to help

with the monitoring and checking. The study

encouraged the school English teachers to

participate in the study voluntarily. Beside their

participation in the study, they had their own

professional and personal commitments and two

months was the agreed time between pre-tests and

post-tests.

1.1 Previous Studies on ESL Students

and Rasch

There were no reported studies conducted, similar to

the current study, from Indonesia or Aceh where

Middle School Students’ behaviour in learning

English as a foreign language was measured, or in

Southeast Asia. The following are previous studies

on ESL students. Smith (2009) examined the

behaviours of EFL Haitian

students

. The

examination included engagement in lessons with

students’ dissimilar levels of English proficiency

(beginning, intermediate, and

advanced

)

during a

Sheltered English Immersion (SEI) English

Language Arts (ELA) literacy

block. The study

resulted in two inferences:

First, support is

needed for English language learners in their

primary language during English instruction, and

second, specific

structural

grouping preferences

should be recognised and supported, along with

exposure to

all

structural groupings. Tillema et al.

(2011) investigated whether students’ response on

offline questionnaire can predict their online meta-

cognitive processing during their writing tasks. The

results showed that the online sequential

disseminations of reading the task and preparation

are unalike for dissimilar degrees of informed

writing styles. Brown & Sachdev (2009) looked at

bilingual identity, vitality, behaviour, and attitudes

of 95 Japanese speakers in United Kingdom. The

findings confirmed the dominance of Japanese in

proficiency and identity. It suggested some

methodical variances in use and attitudes in relation

to context; integrating the three factors related to

identities and vitalities to predict English use and

attitudes; and the use and attitudes of Japanese to

deal with the students’ social contact. Du-Babcock

(2006) analysed topic management strategies and

turn-taking behaviours in the Hong Kong bilingual

environment. The result suggested that different

communication behaviours occurred in tasks

assigned for both Cantonese and English meetings,

indicating that second-language proficiency is likely

a contributing factor that affects the topic

management of Chinese bilinguals when

participating in Cantonese. Farghal & Haggan

(2006) examined compliments behaviour in

bilingual Kuwait college students. With the 632

compliment responses, the result indicated that the

influence of Arabic was very robust over English,

implying the linguistic and culture foundations

rousing the responses.

Differences in Attitude and Behaviour towards Learning English Monolingually and Bilingually: A Rasch Analysis with Banda Aceh Data

161

Research on attitude and behaviour by using

Rasch measurement have been widely conducted

during past decades. Some were reported here.

Leung & Waugh (2010) conducted a study on

attitude towards career counselling of secondary

level students in Hong Kong. The study was

participated by 182 students. The study revealed that

the majority of the students have positive attitude

towards career counselling at the schools. Merrell

(2005) carried out a study on hyperactive and

impulsive behaviour in young children which was

linked to their academic performances by using The

Attention Deficit Hyperactivity Disorder (ADHD)

criteria. The data was taken from 1821 students from

70 schools and was analysed by Rasch analysis. The

study supported previous works and theoretical

points which lead to an issue whether the criteria

was appropriate to use for all ages or not.

2 RESEARCH METHOD

2.1 Data Collection, Procedure, and

Participants

Three hundred and eighty year-7-students from

thirteen public Middle Schools in Banda Aceh

participated in the study. Three hundred and ninety

four students were selected from schools with

bilingual instruction and 386 students were selected

from schools with monolingual instruction. Similar

characteristics were shared between the two groups.

Their English competence was limited, which was

due to the fact that they just started learning English

in their year 7 because English became one of the

compulsory subjects in that year. They spoke both

Bahasa Indonesia and Acehnese inside and outside

the classroom. In the classroom English was not

spoken, but learnt, especially through reading texts

and grammatical drills. Only a few spoke English,

read English, and watched movies as their hobby. At

schools, both of the students were taught using the

same English syllabi and curriculum; similar English

teaching approaches, methods and strategies, and

similar tasks.

The amount of time spent on English lessons at

schools became the only difference between the two

groups. The amount spent on English lessons a week

for bilingual-taught schools were around 55 hours

while the amount spent on English lessons were 38

hours for monolingually-taught schools for a week.

So, the amount of time spent was the major

difference in spite of their similarity in their English

scarcity.

2.2 Instrumentation

To investigate students’ attitude and behaviour

towards their learning English in both schools type

in Banda Aceh’s context, a Questionnaire test was

designed. The questionnaire which consisted of

questions for both attitude and behaviour,

comprising a total of 21 statements, was divided into

four categories: (1) tasks for listening; (2) tasks for

speaking; (3) tasks for reading; and (4) tasks for

writing. Each statement was responded in three

response categories, ‘never or rarely’, ‘some of the

time’ and ‘most of the time’. ‘Never or rarely’ was

scored 0; ‘some of the time’ was scored 1, and ‘most

or all the time’ was scored 2.

2.3 Measurement

This study uses the computer program RUMM2030

(Andrich et al. 2010). It provides statistics and

graphs to support the assertion that a linear

unidimensional measure was made. There are

number of steps to be considered in using this

analysis (see for example Waugh 2003, 2005,

2010b, 2010a). These included, for example but not

limited to, checking the theoretical ordering of the

items against the measured order; checking the

Standardized Fit Residuals; Item-Trait Interaction

(for respondent agreement on the difficulties of the

items); Person Separation Reliability Index,

Individual Item Fit, Response Category Curves,

differential item functioning and appropriate

targeting (by gender and type of English program).

3 RESULT AND DISCUSSION

3.1 Initial Analysis

The original questionnaire involved attitude and

behaviour responses to 21 items, making an

effective scale with 42 items. The items were

ordered theoretically from easy-to-medium-to-hard

on a continuum with the attitude items expected to

be easier than their corresponding behaviour items.

After the analyses, 20 of the 21 attitude items

(item 7 was the exception) were deleted because of

misfit to the measurement model. This was not

consistent with the model used to develop the

questionnaire that was based on many previous

studies where attitude and behaviour were measured

together (see Waugh 2003, 2005, 2010a, 2010b).

The RUMM program does not tell the researcher

why an item doesn’t fit the Rasch Measurement

BELTIC 2018 - 1st Bandung English Language Teaching International Conference

162

Model; just that it doesn’t fit. It was difficult to see

why the attitude items didn’t fit the measurement

model, but the students were in their first year of

middle school and, because they were not strong in

English reading, the misfit may have been primarily

due to their low command of reading and

understanding of English, related to the different

classroom culture in the two groups, in a test

situation. The two groups of students (bilingually-

taught and monolingually taught) did not have

agreement about the difficulties of the items because

of their differences in their command of English,

combined with some differences in culture (living in

an Indonesian culture, and learning English culture

through English lessons that was taught mostly in

Bahasa Indonesia), and that was the substantial

cause of the misfit. Analysing the 21 attitude items

alone also did not produce a linear scale and so the

analysis was continued after deleting the 20 non-

fitting attitude items.

A final analysis showed ten behaviour items

(items 2, 8, 10, 12, 16, 18, 26, 28, 36 and 38) and

one attitude item (item 7), produced a good fit to the

measurement model. Deletion of the attitude item 7

(“I say new words several times in English”) and a

re-analysis with the ten behaviour items produced a

worse fit to the measurement model and so the

attitude item 7 was re-instated. The following

material shows the output from the RUMM program

when a good, unidimensional, linear scale of

Behaviour with respect to Learning English was

created with 11 items for these Acehnese students.

3.2 Output from Final Analysis

3.2.1 Standardised Fit Residuals

To match the measurement model, the Fit Residual

for both items and students should be near 0 and the

standard deviation should near 1. For this study, the

mean for the Fit Residual had 0.195 for items and -0.

306 for persons and the Standard Deviation was

1.112 for items and 1.525 for persons. It means the

data fit the model in which it had a reasonable

reliability of item-student response pattern. Table 1

shows item-person fit to the measurement model for

the Behaviour measure.

Table 1: Overall Fit Statistics for the Behaviour Measure

(N=779, I=11)

ITEM-PERSON INTERACTION

ITEMS PERSONS

Loca-

tion

Fit

Resi-dual

Loca-

tion

Fit

Residual

Mean 0.000 0.195 0.458 -0.306

Standard

Devia-

tion

0.217 1.112 0.972 1.525

Notes:

1. The mean of the item difficulties is constrained to zero

by the measurement model.

2. The fit residuals will approximate a distribution with a

mean near zero and a standard deviation near one,

when the data fit the measurement model.

3.2.2 Dimensionality and Item-Trait

Interaction

In order to determine whether a unidimensionality

trait has been measured, an item-trait interaction chi-

square is needed. The item trait interaction chi-

square for this study was 103.82, df=99, and p=0.35

(see Table 2). This indicated that all along the scale,

the agreement amongst the students about the item

difficulties was good. It shows that the students

agreed as to which items were the hardest, which

were of medium difficulty, and which were the

easiest. This, sequentially, means that, prediction to

each student’s response to each item can use the

person measure (a single parameter for each student)

and the item measure (a single parameter for each

item). The first eigenvalue which was shown from a

major component examination of the residuals was

1.45 which was acceptable, indicating that the data

produced a unidimensional measure.

Table 2: Item-Trait Interaction for Behaviour Scale

Total Item Chi-Square 103.82

Separation Index 0.71

Total Degree of Freedom 99.00

Total Chi-Square Probability 0.35

Cronbach Alpha 0.73

Notes:

1. The Index of Person Separation was good.

2. The item-trait interaction test specified that there was

good agreement among the students on the item

difficulties.

3. All numbers are given to two decimal points because

the errors are only up to two decimal points.

3.2.3 Person Separation Index

A Person Separation Index is an indicator that the

student measures are well separated along the scale

in comparison to the students’ measurement errors.

To have a good measure, the Person Separation

Index should be 0.75 or larger. The Person

Separation Index for this study was 0.71 (see Table

2 above) showing a rational separation of measures

with regard to the errors. Unlike Cronbach Alpha

Differences in Attitude and Behaviour towards Learning English Monolingually and Bilingually: A Rasch Analysis with Banda Aceh Data

163

which is calculated on the data raw scores, Person

Separation Index is calculated on the parameters that

are created by Rasch. For this study, the Cronbach

Alpha was 0.73, which once more shows that the

data had a reasonable consistency.

3.2.4 Individual Item Fit

For the Attitude and Behaviour measure, eleven

items fit the measurement model (see Table 9.3). In

addition, all the standardized residuals fall within the

range -1.2 to +1.1, supporting a reasonable fit to the

measurement model (which usually has to be within

plus or minus two SDs).

Table 3: Locations, Standard Errors, Residuals and Chi-

Squares for Behaviour Items

Ite

Loca-

tion

SE Resi-

dual

DF Chi-

Squar

Proba-

bility

2 -

0.27

0.06

0.47

706.

5.06

9 0.83

7 -

0.17

0.05

0.94

706.

8.52

9 0.48

8 0.1

0.06

0.09

706.

4.40

9 0.88

10 0.1

0.05

0.69

706.

15.0

9 0.01

12 0.3

0.05

0.03

706.

7.22

9 0.61

16 -

0.28

0.05

1.13

706.

14.5

9 0.10

18 0.1

0.05

0.31

706.

5.07

9 0.83

26 0.0

0.05

0.22

706.

6.84

9 0.65

28 0.1

0.05

0.56

706.

4.02

9 0.91

36 -

0.48

0.05

0.22

706.

11.3

9 0.25

38 0.2

0.05

1.07

706.

9.99

9 0.35

Notes:

1. Location refers to the item difficulty in logits (the log

odds of answering the response categories positively).

SE is the standard error in logits.

2. Residual is the difference between the observed and

expected responses.

3. df means degrees of freedom. Probability is based on

the chi-square fit to the measurement model.

3.2.5 Threshold Values

Items thresholds are positions on the scale between

adjacent response categories where the odds are 1:1

that students will respond to a particular item, in

either category. It is expected that the students

would use the thresholds in the way that they were

intended by the researchers and so the thresholds

should be ordered in line with the conceptual

ordering of the response and scoring categories. The

thresholds were ordered in line with the conceptual

ordering of the scoring categories and thus were in

agreement with the intention of the researchers for

this measure (see Table 4).

Table 4: Item Thresholds Uncentralised (Item=11,

Number=779) for Behaviour Measure

Item Item

Location

Thresholds

1 2

2 -.222 -1.805 1.361

7 -.132 -.804 .539

8 .157 -1.188 1.504

10 .124 -1.038 1.287

12 .352 -.623 1.327

16 -.142 -1.144 .858

18 .061 -.935 1.058

26 .092 -.975 1.160

28 .030 -.831 .893

36 -.439 -1.109 .230

38 .117 -.831 1.066

Note: The thresholds are ordered in line with the scoring

categories.

3.2.6 Scoring Category Curve

The RUMM2030 program produces curves of the

scoring categories for each item. The Scoring

Category Curves show the relationship between the

probabilities of scoring in each category. Each item

has three response categories: ‘Never or rarely’

(scored 0); ‘Some of the time’ (scored 1); and ‘Most

or all the time’ (scored 2). The Scoring Category

Curves should show a consistent relationship

between the probability of scoring and the measure

from low to high indicating that the scoring was

done consistently and logically. A Scoring Category

Curve for Item 2 is shown in Figure 1.

Figure 1 (Item 2: “I pay attention to someone

speaking English”) showed that the scoring was

done logically and consistently. When students have

low measures on item 2, then they have a high

probability of obtaining a zero score (the lowest

response); and when they have a medium measure,

they have a medium probability of scoring 1 (the

moderate response); and when they have a high

measure, they have a high probability of scoring 2

(the highest response). The Scoring Category Curves

for the other items were checked and they showed

logical and consistent scoring as well.

BELTIC 2018 - 1st Bandung English Language Teaching International Conference

164

Figure 1: Item Category Curve for Item 2

Note: The blue curve is for a score of 0, the red curve for a score

of 1 and the green curve for a score of 2.

3.2.7 Item Characteristic Curves

The Item Characteristic Curve provides information

on item differentiation between persons and the item

location. A group of students is considered to have

performed well if their values (in the form of black

dots) fit on the ogive curve. The ogive curve is the

expected values for an item against the student

measures (low to high). Take as an example Item 2

(Figure 2). With most of the dots close to the curve,

it is considered a good fit to the measurement model

and shows good discrimination. The Item

Characteristic Curves for the other items were

checked and found to be satisfactory.

Figure 2: Item Characteristic Curve for Item 2

3.2.8 Person-Item Threshold Distribution

(Targeting)

The RUMM2030 program produces item difficulty

on student measure which is also known as targeting

graph. On that graph, the student measures are

placed on the same scale as the item difficulty in a

standard unit. A well-targeted measure is achieved

when the thresholds and student measures are at

about the same range, indicating that the items are

within the capability of the students to answer.

Figure 3 illustrates the targeting graph for the

Behaviour measure. The targeting of the Behaviour

measure items is not as satisfied as estimated

because there were inadequate easy, medium, and

hard items. This was because several original items

misfit the measurement model and therefore were

deleted after the initial analysis in which it showed

that the two types of students disagreed on most item

difficulties.

Figure 3: Person-Item Threshold Distribution for

Behaviour

Figure 4 shows the distribution of item

thresholds and the students on a ‘map’ and this also

shows the restricted range of item thresholds.

Somehow, this targeting problem would have to be

rectified in any future use of the scale for these

students.

Figure 4: Behaviour Map

Note: I0016.2 means threshold 2 for item 16, I0002.1 means

threshold 1 for item 2, and so on.

Differences in Attitude and Behaviour towards Learning English Monolingually and Bilingually: A Rasch Analysis with Banda Aceh Data

165

3.2.9 Differential Item Functioning (DIF) by

Gender

Each of the 11 items of the Behaviour measure

showed no statistically significant Differential Item

Functioning (DIF) by gender (see Figures 5, 6 and 7,

for examples).

Figure 5: Item Characteristic Curves by Gender for

Behaviour Item 26

Note: No statistically significant main effect by gender, F=5.02,

df=19,1, p=0.025.

Figure 6: Item Characteristic Curves by Gender for

Behaviour Item 12

Note: No Statistically significant main effect by gender, F=2.96,

df=19,1, p=0.09.

Figure 7: Item Characteristic Curves by Gender for

Behaviour Item 38

Note: No statistically significant interaction effect by gender,

F=0.19, df=19,1, p=0.66

All the Item Characteristic Curves for the other

items were checked and showed no statistically

significant difference by gender.

3.2.10 DIF such by Type of Language

Instruction

Only one item (Item 7) showed DIF by type where

bilinguals had improved results on the Behaviour

measure compared to monolinguals (F=23.81,

df=19,1, p=0.00000, see Figure 8). Figures 9 and 10

show that there was no DIF for item 2 and item 38.

However, over the Rasch measures for all the 11

items together, bilinguals had a statistically

significantly higher Behaviour measure than

monolinguals (F=20.56, df=1,778, p=0.0000, see

Figure 12).

Figure 8 Item Characteristic Curves by Type of Teaching

Methods (Bilingual v. Monolingual) for Behaviour Item 7

Note: There is a statistically significant main effect by type,

F=23.81, df=19,1, p=0.00000

Bilinguals have a statistically significantly higher

Behaviour measure.

BELTIC 2018 - 1st Bandung English Language Teaching International Conference

166

Figure 9: Item Characteristic Curves by Type of Teaching

Methods (Bilingual v. Monolingual) for Behaviour Item 2

Note: Not statistically significant by type of teaching (F=5.65,

df=19,1, p=0.02)

Figure 10: Item Characteristic Curves by Type of

Teaching Methods (Bilingual v. Monolingual) for

Behaviour Item 38

Note: Not statistically significant by type of teaching F=4.66,

df=19,1, p=0.03

3.2.11 Targeting by Gender and Type of

Language Instruction

For targeting by gender and type of language

instructions (bilingual and monolingual), girls

revealed a statistically significantly improved

Behaviour measure than boys (F= 11.52, df=1,778,

p=0.0007, see Figure 11) and bilingually-taught

students have a statistically significantly better

Behaviour measure than monolingually-taught

students (F= 20.56, df=1,778, p=0.00002, see Figure

12).

Figure 11: Targeting of Behaviour by Gender

Notes:

1. The person measures are on the upper-side of the

graph from low (LHS) to high (RHS).

2. The item difficulties are on the lower-side side from

easy (LHS) to hard (RHS). F= 11.52, df=1,778,

p=0.0007, which is a statistically significant

difference.

Figure 12: Targeting of Behaviour by Type of Language

Instruction

Note: F= 20.56, df=1,778, p=0.00002 which is statistically

significant for bilingual teaching.

3.2.12 Scale of Item Difficulties

Table 5 shows the item wording for Behaviour. The

items have been ordered by difficulty from the

easiest to the most difficult on the linear Rasch-

created scale. The ordering of the items is consistent

with the initial predicted conceptualized order,

supporting the construct validity of the scale.

Table 5: Order of Difficulty of Items on the Linear Scale

Item

No.

Item

Location

Item Statements on the Behaviour

(easy

)

-0.44 I actually like the way my teacher

teaches English writing.

2 -0.22 I actually pay attention to someone

Differences in Attitude and Behaviour towards Learning English Monolingually and Bilingually: A Rasch Analysis with Banda Aceh Data

167

speaking English.

16 -0.14 I actually read carefully words in

English.

7 -0.31 I wish to say new English words

several times.

28 0.01 I actually can understand English better

when I do activities with friends

18 0.06 I actually can read English at home on

my own.

26 0.09 I actually learn more when I study

English in groups.

38 0.12 I actually like English because we use

it in the classroom.

10 0.13 I actually practice English with other

students.

8 0.16 I actually say new English words

several times.

(hard

)

0.35 I actually start conversation in English

with my friends.

Note: Item difficulties (locations) are measured in logits, the log

odds of answering successfully.

The items were ordered from easy to hard on a

linear scale (see Table 5) so that it can be seen which

items are easy and which are hard. The easiest item

involved the students’ preference for the way their

English teachers teach in the classroom, as expected

(item 36 difficulty = -0.44 logits). The hardest item

involved oral skill in ability to initiate speaking in

English with their friends, as expected (item 12

difficulty = +0.35 logits).

4 CONCLUSIONS

This study presents the results of the data analysis

for the process of students’ behaviour in learning

English as a second language in Aceh. Eleven items

from the original 42 items produced a linear,

unidimensional measure (31 items were deleted: 20

attitude items and 11 behaviour items). The Fit

Residual data showed that there was a good

consistency for the item-person response pattern.

The Item-Trait Interaction (dimensionality)

indicated that there was good agreement about the

item difficulties along the scale. The Person

Separation Index indicated that there was good

separation of measures in comparison to errors. All

items fitted the Rasch measurement model. The

threshold values and the Scoring Category Curves

showed that the scoring categories were used

consistently and logically. The Item Characteristic

Curves showed reasonable discrimination. All these

data support the view that a linear, unidimensional

measure of Behaviour was created so that valid

inferences could be made.

However, 31 items, which were consisted of 20

attitude items and 11 behaviour items and which

were initially considered to be conceptually valid,

had to be deleted because of misfit to the

measurement model, apparently because the

bilinguals and the monolinguals did not agree on the

item difficulties.

On the attitude items (see Appendix A for unfit

Attitude items), the disagreement occurred because

bilinguals and monolinguals think differently about

learning English, consisting tasks on listening, tasks

for speaking, tasks for reading, tasks for writing,

student/student relationship, student/teacher

relationship, and common views. This agreement

was related to different views bilingual and

monolinguals had towards their learning English as a

foreign language, especially regarding the attitude

items that previously mentioned. For tasks for

listening, both groups did not reach agreement on

what they thought on paying attention to someone

speaking English, on asking others to speak slowly

or repeat words in English, and on listening to

English songs. Based on observation conducted on

the time of data collection, it showed that

monolingual students did not pay an adequate

attention on listening tasks. It was strongly

associated with possession of inadequate vocabulary

which was considered crucial in order students to be

able to listen to English conversations well. Added

to it was that this lack of vocabulary contributed to

their un-readiness for English listening activities that

took a place in a monolingual English classroom in

Aceh province. Therefore, it was unlikely that the

monolingual students asked others to speak slowly

or repeat words in English. On the other hand,

bilingual students seemed to have better

achievement in all listening tasks that monolinguals

students were lack from. The bilingual students

showed that they thought they liked English

listening tasks. They were aware that they paid good

attention to someone who was speaking English, for

example, their English teacher or other students with

good ability in English speaking. In line with that,

they also believed that they asked others to speak

slowly or repeat words in English in order that they

learned listening better, or on how to pronounce or

how to gasp a meaning of an utterance. The

bilingual students also thought that they liked

listening to English vocabulary, especially for

meaning of words or expressions, or on spellings or

intonations.

Similarly, for other tasks, bilingual students had

better thought towards tasks for speaking, tasks for

readings, and tasks for writing. They showed better

BELTIC 2018 - 1st Bandung English Language Teaching International Conference

168

motivation and dedication on their learning English,

in this case, for English speaking, English reading

and English Writing, due to the fact that they were

considered to possess sufficient English vocabulary

to start with the tasks. Sufficient English vocabulary

was very crucial because students could create

spoken utterances, understand reading passages, or

write something with. With adequate vocabulary,

bilingual students wished to practice English with

other students and to start conversation in English

with their friends. Having sufficient vocabulary

encouraged them to guess the meaning of the

English words in the text, to read carefully words in

English and to read English at home on their own. In

addition to that, having adequate vocabulary

motivated them to look for similar words in their

own language, and to divide English words into

parts that they understand as well as to write to write

feelings in a diary in English.

On the contrary, monolingual students thought

differently due to the limited vocabulary that they

possessed regarding the above tasks. They did not

have enough vocabulary to be engaged in most of

English tasks. Most of them could only know basic

English vocabulary which was insufficient to trigger

English conversation, English reading, and English

writing.

Different agreement also occurred in the way

both groups of students view their relationship with

their classmates, with their English teacher and on

common views. Bilingual students wished to learn

more when they studied English in groups, to

understand English better when doing activities with

friends as well as to lead other friends in doing

English activities. They also wished to learn a lot

from their English teachers, thought that they like

the way their teachers taught English Reading

Comprehension, and English writing. They liked

English because they used it in the classroom,

because it helped them in higher study, and because

it helped them go abroad. On the other hand,

monolingual students seemed to believe these

differently.

Regarding the behaviour items (see Appendix B

for unfit Behaviour items), the disagreement was

because the bilinguals and monolinguals behaved

differently about learning English, consisting with

tasks for listening, tasks for reading, on

student/student relationship, and on common views.

For tasks for listening, monolinguals students did

not ask others to speak slowly or repeat words in

English, nor listen to English songs the way

bilingual students did. Further disagreement was

also shown on tasks for reading, on student/student

relationship, student/teacher relationship, and

common views. For tasks for reading, they guessed

the meaning of the English words in the text and

look for similar words in their own language. For

tasks of writing, they wrote feelings in a diary in

English and divided English words into parts that

they understood. For student/student relationships,

student/teacher relationships, and common views,

they lead other friends in doing English activities,

learned a lot from their English teacher, liked the

way their teacher taught English Reading

Comprehension, and liked English because it helped

them in higher study and because it helped them go

abroad. However, monolingual students did them

differently with the bilinguals.

The reason for this is similar to the way they

responded to the attitude items. Monolingual

students seemed not to have as strong motivation as

bilingual students and to have lack dedication to

learn English, which could be related to their un-

readiness to be engaged in learning English where

sufficient vocabulary was involved. Compared to

monolinguals, bilinguals had English-based

activities at schools. These activities were believed

to increase the students’ motivation to learn English.

Some activities that had been witnessed were,

speech competition, story-telling, and narrative

writing. Students and teacher dedicated some time in

a week to do practices. In order to do the practices

well, some other activities such as book reading, was

involved, especially for the speech competition. In

doing so, the children were lent with some books or

magazines or articles from the Internet to ensure

them well-informed prior to the story telling. On the

contrary, such activities were not conducted at

monolingual schools. When being confirmed to the

teachers, they told that there were some problems

either with the students’ lack participation or with

teachers’ decision by not taking parts on the

competition. In classrooms, bilingual students were

found active and motivated. Most of the question-

answer sessions were alive. Students raised their

hands to ask questions or clarified things. Teachers

explained materials under questions thoroughly.

Those atmosphere, on the other hand, were rarely

seen on the monolingual classrooms. Students

seemed reluctant to raise hand and ask questions or

did clarifications. When questioned the reason why,

some of them told that they did not know what to

ask for clarification because they knew very little. It

was hard for the monolingual students to get

involved into any English activities with their

limited vocabulary.

Differences in Attitude and Behaviour towards Learning English Monolingually and Bilingually: A Rasch Analysis with Banda Aceh Data

169

Due to this disagreement, the 20 attitude items

and 11 behaviour items need to be deleted. The

deletion of these items caused a targeting problem,

where the final scale had insufficient items across

the whole difficulty range of the student measures.

Thus this problem needs to be investigated further so

that there are sufficient easy, medium and hard items

to cover the full range of student measures in any

future use of this scale.

The Item Characteristic Curves showed that the

large majority of the items had no statistically

significant differential item functioning (DIF) by

gender and by type of language of instruction

(bilingually-taught and monolingually-taught).

However, over the Rasch measure for all 11 items

together, bilinguals had statistically significantly

better Behaviour measure than monolinguals and

females had a statistically significantly better

Behaviour measure than males.

ACKNOWLEDGEMENTS

The author (Khairiah Syahabuddin) thanks the

Australian Endeavour Scholarship (Department of

Education, Employment and Workplace Relation,

DEEWR) for funding her doctorate study at the

Edith Cowan University, Perth, Australia, during

which she conducted this research. Her thank also

goes to Professor Russell Waugh, her supervisor,

without whom she could not figure out herself with

her emerging knowledge of Rasch Measurement.

Her thank is also dedicated to her family who always

supports and prays for her.

REFERENCES

Smith, J., 1998. The book, The publishing company.

London, 2

edition.

Andrich, D., Sheridan, B. & Luo, G., 2010. Rasch

unidimensional measurement model

(RUMM2030) computer program.

Brown, I. & Sachdev, I., 2009. Journal of

Multilingual and Bilingual behaviour,

attitudes, identity and vitality: Some data from

Japanese speakers in London, UK. , 30(July,

2009), pp.37–41.

Du-Babcock, B., 2006. An Analysis of Topic

Management Strategies and Turn-Taking

Behavior in the Hong Kong Bilingual

Environment: The Impact of Culture and

Language Use. , 43, pp.21–42.

Farghal, M. & Haggan, M., 2006. Compliment

Behaviour in Bilingual Kuwaiti College

Students. , 9(1), pp.94–118.

Leung, Thomas Kin Man; Waugh, R., 2010. A

Rasch Measure of Career Decision-Making

Self-Efficacy. In Applications of Rasch

Measurement in Education. pp. 143–154.

Merrell, C., 2005. Rasch Analysis of Inattentive,

Hyperactive and impulsive Behaviour in

Young Children and the Link with Academic

Achievement. Journal of Applied

Measurement, 6(1), pp.1–18.

Smith, M., 2009. Academically engaged behaviors

during literacy instruction for three Haitian

English Language Learners in a Sheltered

English Immersion First Grade Classroom.

Tillema, M. et al., 2011. Relating self reports of

writing behaviour and online task execution

using a temporal model. , 6, pp.229–253.

Waugh, R.F., 2010a. Applications of Rasch

measurement in education, New York: Nova

Science Publishers.

Waugh, R.F., 2005. Frontiers in educational

psychology, New York: Nova Science

Publishers.

Waugh, R.F., 2003. On the forefront of educational

psychology, New York: Nova Science

Publishers.

Waugh, R.F., 2010b. Specialized Rasch measures

applied at the forefront of education, New

York: Nova Science Publishers.

BELTIC 2018 - 1st Bandung English Language Teaching International Conference

170