3.3 Instruments
3.3.2 The Han-Lin iEnglish Book 1
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
36
3.3.2 The Han-Lin iEnglish Book 1
At the end of every academic year, English teachers are required to choose an appropriate textbook version suitable for the junior high school students' cognitive levels. The members of English Teaching Committee in this school voted the Han-Lin iEnglish Book 1 for seven graders at the subjects’ school in 2013 for several reasons.
First, the textbook was edited based on the Nine-Year Integrated English Curriculum Guidelines of Elementary and Junior High Schools (Ministry of Education, Republic of China, 2008). Second, the publisher of the textbook could provide many useful supplementary teaching materials, such as posters, CDs and newspaper. Third, the content of the textbook could reflect the authenticity and appropriateness of the language, and the trends of the latest topics. Fourth, the four skills are integrated in a balanced way and there was no cultural bias in it. Fifth, the format of the textbook was characterized by the clarity of typesetting, good quality and clarity of illustration and editing. According to these practical criteria, the Han-Lin iEnglish was chosen as the instructional material for the seven graders in 2013 academic year.
There are nine units in the textbook, starting from Starter to Unit 8. Each unit at least covers one grammar focus with sentence patterns. The presentation of the sentence pattern in each unit usually follows the following structure: tables of examples first, and then some grammar practices such as gap-fills or making sentences with picture cues (Figure 3).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
37
Figure 3. The presentation of the sentence pattern in Han-Lin iEnglish Book 1 3.3.3 Tiered Task--the Three Differentiated Worksheets
Heaton (1990) mentions that assessments designed by the teachers are the most useful because teachers understand their students’ strengths and weaknesses as well as the skills and language areas needed to be focused on. Classroom assessment, such as writing worksheets, can help teachers assess learners’ performance, enhance teachers’
effectiveness and benefit students (Hsieh, 2011). Additionally, Gower and Walters (1983) pointed out four main approaches for teachers to use self-designed worksheets:
1) to copy the materials which aren’t available to students, 2) to make cards for communication activities, 3) to make the teacher’s own exercises and 4) to adapt published materials.
In this study, the last two approaches were adopted. The researcher adapted the grammar materials in the textbook and designed three differentiated worksheets for grammar exercise based on the same grammar points of the unit. In other words, the participants who were grouped by their performance on the pretest would be assigned to different tiered tasks.
Tomlinson's tiered tasks were used as the framework to construct the
‧
differentiated worksheets in this study. Tomlinson (1999) described a "tiered task" as a differentiated strategy that addresses a particular standard, key concept and
generalization, but allows several pathways for students to arrive at an understanding of these components based on students' interests, readiness, or learning profiles.
Among the important reasons for choosing Tomlinson's work are that this work addresses the needs of all learners at different levels. In this study, the focus is not only on advanced learners, but also on average and below anticipated level learners.
This work emphasizes effective instruction in heterogeneous settings. Because a typical EFL classroom at junior high school in Taiwan is comprised with
mixed-ability students, Tomlinson's work fits the population of this study.
A whole set of tiered tasks presented three tasks for learners to produce different linguistic output. In this study, Task A for LPL was multiple-choice questions with three possible answers on the worksheet (Appendix D). Task B for IPL was gap-fill questions which were slightly different from Task A, in that the subjects had to write down answers on their own, without possible choices provided to them (Appendix E).
Task C included 10 sentences-making exercises with picture cues. Task C provided no extra support except picture cues in order to challenge HPL (Appendix F).
3.3.4 Focus Group Interview
A group interview is essentially a qualitative data gathering technique (Denzin &
Lincoln, 2011). Evidence from focus group interviews suggests that attitudes and perceptions associated with concepts are developed in part by interaction with other people (Krueger & Casey, 2009). Blumer (1969) commented that a group interview has the potential to bring a small individuals together as a discussion and resource group and works better than a one-on-one interview in promoting self-disclosure among participants. The intention of using focus group interview in this study was to
‧
understand participants’ perceptions, feelings, attitudes and motivation towards DI.
Edmunds (2000) recommends a full-sized focus group with eight to ten participants when the participants are teens. This size allows interaction while not requiring each individual participant to constantly speak which will create a more comfortable environment and will be more conducive to a successful discussion. Therefore, the researcher conducted three focus group interviews with the interviewees from LPL, IPL and HPL in the experimental group respectively in order to gather data for research question three.
These questions were used as a guide for the interview process (Appendix G).
Based on Hatch (2002), a protocol is only a guide to help you prepare for an interview.
Because a researcher never knows what data a participant will provide, the researcher should allow the participants' responses to guide the interview process. Since this was the basis of the qualitative portion of this research, these questions were a valid way to start a interview. The questions are listed as follows and translated into Chinese version for the subjects. Questions 1 and 2 required the interviewees to describe the process of grammar learning, state their reasons why they can/can't keep up with the process and . Question 3 asked the interviewees whether they liked the process or not.
Questions 4 and 5 focused on realizing the interviewees' opinions about the learning materials. Question 6 collected the data of the participants' learning difficulties in English and question 7 asked if DI was helpful to solve the difficulties they mentioned in question 6. The last question encouraged the interviewees to share their thoughts with others.
3.4 Design and Procedures
A lesson plan was proposed to describe the design for the differentiated instruction. There were three stages in the lesson plan as follows.
‧
Stage 1: The instruction of grammar rules
In this stage, the instructor explained the usage of grammar rules to the whole class. And the a whole set of tiered tasks were assigned to the three groups of learners.
These tasks allowed different proficiency learners to produce different linguistic output.
Stage 2: Practice for grammar rules
After the arrangement of tiered task, the participants were required to finish the task in their flexible grouping setting. While the students were engaging in their tasks, the instructor worked around within the groups to give additional instruction to the slow learners.
Stage 3: check answers and assign homework
In the last stage, the instructor checked the answers of the differentiated
worksheets with LPL, IPL and HPL respectively. The students who finished the task could review or preview the lesson by themselves. In the end of the class, the instructor quickly reviewed the grammar rules and assigned homework.
In addition to the tiered task, the process for the participants to complete the tasks was differentiated as well. The researcher applied the framework of flexible grouping to ensure the learning process was based on DI. Chapman and King (2007) mentioned that flexible grouping gives students opportunities to learn information in a total class, alone, with a partner or with a small group. Other alternatives include random groupings, peer-to-peer tutoring, multiage teams, or cooperative learning situations. According to Ford (2005), the whole class instruction often meant many students didn't learn the appropriate content for their levels. Thus, in this current study, the researcher adapted Ford (2005) model: Grouping Without Tracking
(Differentiating by Levels of Support) to effectively address the diverse needs of
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
41
learners. The three groups of students were arranged to three group designs respectively. LPL were put in a small group where they can discuss with more partners and the teacher can efficiently provide direct support to a group of low achievers. IPL who needed less support from the teacher were put in pairs and they could share with their partner. HPL who had the ability to learn independently were arranged to complete the task individually. Table 3.4 illustrates the content of 3 tiered tasks and flexible grouping for LPL, IPL and HPL.
Table 3.4
Content of 3 Tiered Tasks and Flexible Grouping
participants tiered tasks flexible grouping
LPL Task A: 10 multiple-choice items small group
IPL Task B: 10 gap-fill pairs
HPL Task C: 10 sentences-making with picture cues individual
Table 3.5 states the design of the experimental group and the control group during the whole study.
Table 3.5
The Operation of the Experimental Group and the Control Group
the experimental group the control group
learning content three differentiated worksheets exercises of the textbook learning process flexible grouping learning traditional lecture method
Besides, the procedure of the study was implemented through eight phases. Each phase was stated in detail as follows (Table 3.6).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
42
Table 3.6
The Procedure of the Main Study
Phase Time Content
Phase 1 2013.8 The researcher designed the two English Grammar Tests, the pretest and the posttest.
Phase 2 9.2-9.12 After piloting the pre-and posttests, the researcher revised them according to the result of item analysis.
Phase 3 9.13-9.15 All the subjects took the pretest in the formal study. The scores of the pretest were utilized to divided the participants into three subgroups.
Phase 4 9.16-9.20 Two classes of the subjects were randomly assigned to be the experimental group or the control group.
Phase 5 9.16-11.29 The researcher administered DI for 11 weeks to the experimental group.
Phase 6 12.2-12.6 All the subjects took the posttest in the 12th week of the semester. The researcher collected the scores of the posttest for data analysis.
Phase 7 12.9-12.13 The subjects of the experimental group received focus group interview in the 13th week.
Phase 8 12.16- With the data, the researcher made comparisons of the subjects scores between before and after DI. Meanwhile, the research transcribed the focus group interview to do the content analysis.
‧
The statistical package SPSS was used to compute the data collected from the pre- and post-tests to answer the research questions listed earlier in chapter 1.
This study adopted a quasi-experimental design. Both qualitative and
quantitative analyses were used to answer research questions. The quasi-experimental design is a common alternative to true experimental designs in the evaluation of education programs when, as in this study, random assignment is not possible or practical. Due to the nonequivalent-groups design, analysis included adjusting the pretest scores for measurement error. Otherwise, one-way analysis of covariance (ANCOVA) model may contain bias and may not present the true treatment effect.
To answer the first research question, after adjusting the pretest mean scores, ANCOVA was implemented to compare the adjusted mean scores of the posttest to check if there is a significant difference between the experimental group and the control group.
To answer the second research question which aims at evaluating the effects of differentiated and non-differentiated instruction on high, intermediate and low proficiency learners respectively, descriptive statistics were stated first in order to provide information related to central tendency and dispersion Then ANCOVA was performed to examine if learners of different proficiency level benefit from DI differently.
As for the third question, a focus group interview was employed to explore the participants’ thoughts toward DI. The purposeful sample of participants in the focus group was to better understand the implementation of DI. Data collected from the focus group required that careful notations be made to ensure coding and emerging themes were communicated accurately based on the participants' conversation. The
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
44
note-based content analysis was used to analyze the focus group interviews. The raw data used for the note-based content analysis relied primarily on filed notes, which were based on observations and comments in the interviews, a debriefing session and summary comments at the conclusion of each interview (Krueger & Casey, 2009). In order for analysis to be verifiable, the data stream began with field notes and
recordings that were taken during the interviews, continued with the oral summary of key points during the interviews, and also included the electronic recording with the possibility of a translation of the interviews (Krueger & Casey, 2009).
After the interview, a peer expert with a background in qualitative research was invited to review the frequency of counting and data interpretation in order to check the researcher's accuracy in this analysis. The coding discrepancies were discussed and resolved by the third coder to ensure the trustworthiness.
‧
This chapter presents the result derived from the present study and a summary of the findings. The present study investigated the effects of differentiated instruction on junior high school students' grammar learning in Taiwan. The main issue was
explored in three areas:
1. Are both “differentiated instruction” and “non-differentiated instruction”
equally beneficial for students on their grammar learning?
2. Did high-, intermediate- and low proficiency learners benefit to different extent from "differentiated instruction" and "non-differentiated instruction"
respectively?
3. What were the learners’ opinions after they had experienced differentiated instruction?
4.1 Comparisons of the Participants' Performance in the Pretest and Posttest Descriptive Statistics for the pretest and posttest results are presented in Table 4.1. Range of scores as well as observed scores on the pre and post measures are shown. All distribution were are plotted and scanned for kurtosis and skewness in Figure 1.
Table 4.1
Descriptive Statistics for the Experimental Group and the Control Group
N Mean Range SD
Group pre post pre post pre post pre post
EG 27 27 38.29 40.70 12-56 6-56 10.90 12.34
CG 27 27 36.14 34.07 12-54 2-58 13.54 17.02
Note: EG: Experimental Group; CG: Control Group
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
46
The mean raw scores of pretest and posttest are graphically displayed in Figure 1 below. Visual analysis of the scatter plot provides evidence that the two groups do not appear to be equivalent in terms of their scores. Their test scores ranged from 12 to 56 before the study and ranged from 2 to 58 after the 11-week intervention.
Figure 4. Bivariate distribution of pretest and posttest for the experimental and control groups
Each distribution was found to skew negatively, indicating a higher proportion of subjects fell above the mean.
With the aim of examining whether or not the grammar knowledge of the experimental group and the control group differed before DI, the Independent Samples t-test was run to analyze the scores on the pretest. The results are displayed in Table 4.2.
‧
T-test Results of the Pretest Performance
Group N Mean SD t value df p(2 tailed)
EG 27 38.29 10.90 .642 52 .524
CG 27 36.14 13.54
The t value is 0.642 and the probability level is 0.524, confirming that there was no significant difference between the control group and the experimental group at the outset of the experiment.
Table 4.3 shows ANCOVA results for the posttest using adjusted the pretest scores. Results show that for the posttest, the adjusted posttest mean of the
experimental group (M=39.51) was significantly different from the adjusted posttest mean of the control group (M=35.26), F1,52=6.48, p=.01. Statistically speaking, students in the experimental group improved their performances after receiving DI.
Table 4.3
ANCOVA Results of the Posttest Adjusted means
EG CG df F P
Partial Eta squareda
posttest 39.51 35.26 1 6.48 .01 .113
a According to Cohen (1997), Partial eta squared may be interpreted as .01= small effect size, .06= medium effect size, .14=large effect size.
4.2 HPL Performance in the Pretest and Posttest
In order to see whether HPL in the experimental group improved their grammar learning under the researcher's DI during the eleven weeks, the HPLs' performances
‧
on posttest were compared with ANCOVA while controlling for the covariate-pretest scores. Table 4.4 states the raw scores of pretest and posttest for each group and the ANCOVA results are presented in Table 4.5. Results indicated the adjusted posttest mean of the experimental group for HPL is 51.27 and of the control group is 48.97 but the difference of HPL between the two groups did not differ significantly (p=. 38).
Table 4.4
Mean Raw Scores and Standard Deviations for HPL
HPL in EG HPL in CG
Item N M SD N M SD
pretest 8 49.25 3.69 8 49.75 3.91
posttest 8 51.12 4.29 8 49.12 6.33
Table 4.5
ANCOVA Results for HPL on the Posttest Adjusted means
In addition, part of the second research question can also be discussed according to the findings in the Independent Samples t-test for HPLs' scores on each subsection of the posttest. The mean raw scores for each subsection and the comparison with t-test are listed in Table 4.6. The t-test yielded no significant difference in the subsection of multiple choice (p=.149). As for fill-in blank, there was also no
significant difference found between the two groups (p=.483). And the comparison of
‧
the sentence making part between the HPLs in CG and EG did not reveal significant difference (p=.300).
Table 4.6
T-test results for HPL Performances on Each Subsection of the Posttest
Group N Mean SD t value df p
The IPL performance in the two tests are displayed in Table 4.7. The scores are analyzed with ANCOVA as well, and the results are presented in Table 4.8. The probability level (p) for IPLs' performance on the posttest is 0.075, indicating that IPL in the experimental group did not perform significant better than those in the control group on the posttest.
Table 4.7
Mean Raw Scores and Standard Deviations for IPL
IPL in EG IPL in CG
Item N M SD N M SD
pretest 10 40.2 1.98 10 40.4 3.23
posttest 10 44.5 5.27 10 39.1 9.50
‧
Aiming at explore the effect of DI on each subsection, the researcher analyzed the scores of different subsections with Independent t-test. The results are showed in Table 4.9.
Table 4.9
T-test Results for IPL Performances on Each Subsection of the Posttest
Group N Mean SD t value df p
The p level of multiple choice section is 0.14, which did not reach the 0.05 level.
It suggested that the performances of IPL in the experimental group did not
significantly differ than those in the control group. As for fill-in blank and sentence making parts, the p values are 0.415 and 0.223 respectively, which both did not reveal
‧
significant differences between IPL in EG and those in CG on these two sections.
4.4 LPL Performance in the Pretest and Posttest
As for LPL performance in the pretest and posttest, the results are summarized in Table 4.10. ANCOVA was employed to analyze the scores of the posttest, and the results are presented in Table 4.11. It showed that the LPL performance of the
experimental group was not significant different than those of the control group after DI (F=1.284, p=0.275), revealing that DI did not help LPL improve their grammar learning significantly.
Table 4.10
Mean Raw Scores and Standard Deviations for LPL
LPL in EG LPL in CG
Item N M SD N M SD
Pretest 9 26.44 9.36 9 19.33 6.70
posttest 9 27.22 10.79 9 15.11 11.95
Table 4.11
ANCOVA Results for LPLs on the Posttest Adjusted means
Table 4.12 indicated the results of the t-test that LPL in the experimental group perform better than those in the control group on the multiple choice section
statistically (p=0.033). However, in the rest two sections: fill-in blank and sentence
‧
making parts, the results of LPLs' performance did not present there were significant differences existed between EG and CG.
Table 4.12
T-test Results for LPL Performances on Each Subsection of the Posttest
T-test Results for LPL Performances on Each Subsection of the Posttest