國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
generalization test, the questionnaire and interviews were analyzed. The procedure is presented in Figure 3.1.
Standardized Word Recognition Test
Generalization Pre-test and Questionnaire
Letters of Approval for Parents
Stage 1: Participants Selection and Pre-training Assessment
(Experimental Group) Rime Analogy Training
Figure 3.1 The Procedure for Conducting the Formal Study
(Control Group) Phonics Training Stage 2: Training
Interview
Stage 3: Post-training Assessment
Generalization Post-test and Questionnaire
Data Analysis Stage 4: Analysis
‧
Participants in the main study were two10 classes, each composed of approximately thirty sixth graders, aged around 12, in a public elementary school in Tao Yuan County.
They had taken formal English classes once a week from the second grade to the fourth grade, and twice a week in fifth grade. The rules of digraphs, long vowels and
consonant blends had not been formally introduced in the phonics section of their textbooks by the end of the fifth grade.
To ensure the two classes were of the same overall English proficiency level, a standardized English word recognition test11 was administered one on one by the researcher and two other English teachers of the participants’ school before the formal study.12 The test scores of the two classes are listed in Appendixes D-1 and D-2. The statistical results summarized in Table 3.3 indicate that the proficiency levels of the two classes could be considered similar (t = 0.094, p > .05).
ithin
10 The researcher only taught three classes of sixth graders. Based on the three classes’ past overall English performances (i.e., listening, reading, and writing abilities) in school, two similar classes were selected as participants.
11 Although the test mainly assesses examinee’s reading and speaking skills, to assess their overall
proficiency level, other language skills were checked earlier by their overall English performances in school (see the previous footnote).
12 To make sure there was no learning effect during the screening process, the process was completed w two weeks. I’m indebted to Wu, Y. C. (2007) for calling to my attention that administering the test after school might raise a safety problem. Considering the feasibility of conducting the test at school, the researcher asked two colleagues (English teachers) for assistance to speed up the process. I’m grateful for their generous help.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 3.3 Results of the Standardized English Word Recognition Test Between Groups
Group Mean SD t p
Class A (N=29) 47.10 55.56 0.094 0.925
Class B (N=30) 45.87 45.10
Note. Total scores =200
Next, the two classes were given the one-on-one generalization pre-test. To avoid ceiling or flooring effects, 4 students from class A and 5 students from class B were excluded from the study based on their generalization pre-test scores. Moreover, statistical results revealed the two classes were at the same level of decoding abilities (t = -0.273, p > .05) (see Table 3.4).
Table 3.4 Results of the Generalization Pre-test Between Groups
Group (N =25/group) Mean SD t p
Experimental 29.16 23.45 -0.273 0.786
Control 30.92 22.03
Note. Total score =75
Finally, the two groups were randomly assigned to an experimental (RA) group and a control (phonics) group. The numbers of males and females are basically equally distributed (see Table 3.5).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 3.5 Gender and Number of Students in Both Groups
Group Male Female Total
Experimental 12 13 25
Control 12 13 25
Instruments
Testing Materials Standardized English Word Recognition Test
A standardized English word recognition test, designed for Taiwanese children from the third grade to the ninth grade, was implemented one-on-one to examine children’s English word recognition ability (Hong et al. 2006). The purpose of this test was to measure participants’ English proficiency level only.
According to the instruction manual (Hong et al. 2006), one hundred words were chosen based on the 2000 words in the Grade 1-9 Curriculum promulgated by the MOE, a word list from Cobuild English Dictionary for Advanced Learners (CEDAL), and a word list from the Brown Corpus word bank. The examinees were asked to read each word out and give a meaning for each word. All of the responses were tape-recorded for later scoring purposes. Each item received 2 points, one for pronunciation and the other for meaning. Maximum scores for pronunciation and meaning were 100 points each. The
reliability and validity were treated.13 The administration time of the test varied a lot
13 According to the instruction manual for the standardized English word recognition test, the reliability of
‧
from student to student; it ranged from less than one minute for those who completely gave up to those high proficiency students who finished in around 10 minutes for the 200 items (100 for meaning; 100 for pronunciation). The test scores are listed in Appendix D-1 and D-2.
Generalization Test
The generalization test was a modification of the generalization test from Bruck and Treiman’s (1992) experiment. Pre- and post-tests of the generalization test were the same, but with different purposes. The pre-test was used to ascertain whether the two groups were homogeneous in decoding abilities before training, while the post-test was an instrument used to assess students’ progress in the two groups.
The test items were 2514 nonwords derived from the clue words (see Appendix E), which were words from their current or previous textbooks.15 The reason for the use of nonwords as test words is that nonword reading is usually used for looking into learners’
decoding abilities (Chu, 2002) and the researcher could ensure that students did not learn the nonwords on their own time other than during the training period. Care was taken in
this test was .994 in a sample of 464 students of the 5th and 6th grades. To raise the face validity, the test items were reviewed by English teaching experts, experienced English teachers from elementary school or junior high school teachers, and experts in test developing.
14 All the test words were derived from the clue word list, corresponding one to one. There were 20 clue words taught in the present study, but it was hoped that both groups could apply the training strategies to untrained words, too. Therefore, 5 untrained clue words were added. Accordingly, 5 test words were generated. In addition, all the words in the test were examined by a professor in phonology, and an English teacher from the participants’ school. The reliability (Cronbach’sα) was 0.965.
15 The textbooks participants used are: Wonderland (Red): Kid Castle Internet Technologies Ltd.; Welcome 1: Pearson Education Taiwan Ltd.; Melody Smart 2,3,4: Melody Publishing Co., Ltd; English 6 HERE WE GO: Pearson Education Taiwan Ltd.
‧
constructing the 25 nonwords so that these words were not biased towards either group.
That is to say, these words could be decoded by either phonics rules or rime analogy. In creating the nonwords, the researcher selected the graphemes carefully. To begin with, not all single consonants were adopted.16 Then, the consonant digraphs and vowel digraphs were chosen, based on their textbooks. Phonemes included in the test are displayed in Table 3.6.
Table 3.6 Graphemes Included in the Nonwords
Single consonant b, d, f, g, h, k, l, m, n, p, s, t, x Consonant digraphs ch, ck, ng, sh, th
Single vowel letters a, e, i, o, u
Vowel digraphs ai, ea, ee, oa
All the test words were printed out on A4-sized paper. At the beginning of the test, each participant was given instruction in Chinese and required to read aloud the test words one by one (see Appendix A). Their answers were recorded for later scoring.
The scoring criterion followed17 that of nonsense word fluency scoring in DIBELS (Dynamic Indicators of Basic Early Literacy Skills) to count every correct phoneme (e.g.,
16 The rest of the consonants not appearing in the nonwords are c, j, q, r, v, w, y and z. It was found in the pre-pilot that students were able to utilize the strategy but mispronounced the words or even gave up reading the words because they were hampered by blending more difficult phonemes such as v, w and z with rimes. In light of this, to focus on whether participants could use the strategy, not all single consonants were chosen. Confined to those easier consonant sounds listed in Table 3.6, only 11 real nonwords were made up in the end. The remaining 14 words were low frequency real words or archaic words from the Oxford English Dictionary (OED, 1989) (see Appendix E).
17 I am grateful to Wu, Y. C. for calling my attention to counting every phoneme in the verbal production of students and graciously suggesting following the DIBELS scoring, as she did in Wu (2007).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
/k/, /o/, /t/ in “coat”) articulated as 1 point (Good & Kaminski, 2002). For each CVC item, a participant was given 3 points for pronouncing all 3 phonemes correctly. The maximum score on the 25 items was 75. The time participants took for taking the test ranged from 46 seconds to 3’35’’ for the pre-test and 50 seconds to 6’27’’ for the post-test.
Questionnaire on Attitudes toward Reading English Words
Another interest of the present study was to see whether rime analogy training could positively change participants’ English word reading attitudes. A questionnaire
measuring participants’ attitudes toward reading English words was administered before and after the training for both the experimental and the control group (see Appendix B).
This questionnaire was modified from the Reader Self-Perception Scale (Henk & Melnick, 1995) and Chen’s (2003) questionnaire of students’ English word recognition confidence.
The 11 items represented subscales of progress, observational comparison, social feedback, and physiological states (see Table 3.7).18 Each item was rated by 4-point Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). Total scores of the questionnaire were obtained by adding up the scores on the 11 items.
18 Originally, 14 questions were piloted on the 29 students of the third class. The figure suggested that 3 items whose correlations were below 0.1 should be deleted to raise the overall reliability (Cronbach’sα) to 0.853. The 3 questions are: Q5 (PR): When I read English words, I need to consult with others. Q7 (SF): I feel that my classmates love listening to me reading English words. Q8 (PR): It takes me a lot of efforts to read out English words.
‧
Table 3.7 Subscales of the Questionnaire
Subscales Questions Q3 I know how to read out English words.
Q8 My English word reading speed is fast.
PR (Progress)
Q10 My English word reading ability is good.
Q2 I find that my English word reading accuracy is better than my classmates’.
Q4 I am able to read out more English words than my classmates.
OC (Observational
Comparison)
Q6 Reading out English words helps me learn new words faster than my classmates.
SF (Social Feedback)
Q11 My teacher thinks my English word reading ability is good.
Q1 I think reading out English words is interesting.
Q5 I enjoy reading aloud English words.
Q7 I love trying to read out English words.
PS (Physiological
States)
Q9 Reading out English words is easy for me.
Interview
The interview was designed to probe deeply into the thinking process and learning difficulties of the participants in both groups.19 Therefore, following Chang’s (2009) interviewee selection, interviewees were participants either showing positive attitudes but not performing very well comparatively on the generalization post-test, or regressing on attitudes after the training. The questions asked in the interview are presented and discussed in Table 3.8.
19 I’m grateful to Prof. Yu’s suggestion of including the control group as interviewees to compare with the experimental group.
‧
Table 3.8. Design of the Interview Questions
Questions Purposes
Q1.You pronounced “XXX” on the test. Did you try using the teacher’s method?
To know participants’
thinking process.
Q2.You did not say the word last time. Why not? To know participants’
thinking process. (Optional, asked only if the student failed to pronounce the word.)
Q3.Do you think the twenty-session training is enough? To know whether
difficulties result from lack of practice.
Q4. What do you think of the teacher’s teaching method?
Options are provided below if participants have no response or the response is vague.
1) clear
Q1 and Q2 were designed based on the assumption that even though clue word prompts are found to play a crucial facilitative role in the pronunciation of target words (Savage, 1997), participants in this study as EFL learners might still have difficulties retrieving clue words on their own. The researcher used each interviewee’s 3 to 5 incorrect answers in the generalization post-test as examples to probe participants’
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
thinking process and learning difficulties.
Q3 and 4 were developed based on Hsu’s (2003) study, which aimed to promote elementary school students’ English learning attitudes, ability in phonemic awareness, word recognition, and spelling. In Hsu’s study, most interviewees reflected that they felt the teacher taught too fast, too much, and that it was confusing; and the researcher Hsu pointed out that the training duration was not enough for some students. These two problems were inevitable in whole class teaching because the teaching is usually more suitable for the mid-level students; however, it is still worth exploring individual perceptions, especially in those ones who were detected to have learning difficulties.
The wording of the interview generally followed Chang’s interview design (2009), but was modified according to students’ responses in the pilot study.
Teaching Materials and Training
Teaching Materials
The teaching materials, 100 English words20, for both groups were identical but some of them were taught in different sequences throughout the 20 training sessions. The teaching materials were comprised of two parts. One was the clue words and the other was the training words. The clue words were the same for both groups and were ordered from simple to complex (see Appendix F-1, F-2). However, the training words were
20The teaching materials were matched as closely with the clue words as possible, leaving 9 of the 100 words to be nonwords but rhymed with the clue words.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
different for the 2 groups: in the experimental group, all the training words shared the same rime with the clue word, whereas the training words were randomly arranged in the control group. To be more specific, for the experimental group, the teaching materials were 2021 sets of phonograms (i.e., word families). In each session, one set of
phonograms, including 1 clue word (e.g., king) and four rhyming words (e.g., bing, ding,
sing, ring) was taught (Appendix F-1). The rhyming words functioned as practice words, since it was found that vocabulary size was related to the use of orthographic rime
correspondences (Bowey & Underwood, 1996). However, Taiwanese elementary school students might not have a large vocabulary size and by the researcher’s teaching
experience, the practice of rhyming words can reinforce the use of analogy. Therefore, 4 rhyming words were added. For the control group in each session, though the same clue word (e.g., king) was taught as in the experimental group, 4 different words bearing no orthographic similarities (e.g, set, dub, lash, moat) were used as training words.
The structure of all the words was in CVC patterns.22 Half of the 100 words contained digraphs, on the grounds that digraphs also frequently appeared in the participants’ textbooks. Table 3.9 depicts the digraphs used in the 50 words.
21 The 20 rime patterns were chosen either from the participants’ textbook word bank or from the 2000 word list recommended by MOE (2006), with an aim that, in the future, participants could learn new words from the 2000 word list more easily using the rime analogy strategy (see Appendix G).
22 According to Morris (2008), the most fundamental one-syllable pattern is the short vowel pattern, a short vowel preceded and succeeded by consonants. The CVC pattern in the present study refers to simple CVC words (e.g., cat) and more complex variations of the CVC pattern containing vowel or consonant digraphs (e.g., goat, beach).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 3.9 Digraphs Included in the Teaching Materials Consonant digraphs ck, sh, th, ng, ch
Vowel digraphs ea, oa, ai
Training
Both groups received 10 minutes of training for each session, 2 sessions a week. In other words, the total training time was 200 minutes in both groups. The training took place in the first 10 minutes in the participants’ regular English classes of 40 minutes.
The difference in training of the 2 groups was in teaching the 4 training words in the presentation and practice part. The following are examples from one teaching process (e.g., the 15th session, see Appendix H-1, H-2).
In the experimental group, there were 3 teaching procedures: warm-up, presentation and practice, and production. The warm-up stage23 was a quick review of the 5 words taught from the last session. Next, in the presentation and practice section, the teacher started by refreshing students with the clue word king by showing them the page
containing king in their previous textbook and then putting on the board two flash cards, one with the word and the other with the picture of a king. Subsequently, the teacher elicited the participants to read out the new word bing using rime analogy strategy by placing the flash card of bing under the flash card of king. In the same way, practices
23 Although session one had no warm up stage for reviewing words from a previous time, the researcher spent a similar amount of time explaining the up-coming procedure for the first time. So the total training time of the present study was still considered 200 minutes.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
were carried out for the remaining 3 pairs (king-ding, king-sing, king-ring). To reinforce the participants’ familiarity with RA, the teacher went over the same practice to teach participants the 4 practice words again. Lastly, in the production section, the teacher had students copy the rime of the 4 pairs (i.e., king-bing, king-ding, king-sing, king-ring) on their worksheets.24 Five group leaders25 read the four pairs to the teacher first and then went back to their groups to check if the other members in his or her group could do the same practices. If the students could not say the pairs correctly, their group leader modeled it for them and asked them to repeat. The practice was completed by ticking a checklist (see appendix H-1, H-2) for the group members.
Figure3.2 Teaching Setting for the Experimental Group (Rime Analogy Group)
Likewise, in the control group, the same procedure was carried out but this group differed in the approach to teach the 4 training words in the presentation and practice
24 The format of the worksheet is shown in the production procedure in each session (see Appendix H-1, H-2). Every participant was given one worksheet from the researcher and asked to keep it in the file folder so the researcher could check their records at the completion of the present study.
25 The five group leaders were selected from each class based on their final exam scores in fifth grade and their good personalities: being willing to help others.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
section.26 To illustrate, the teacher showed participants the word set, and placed the strip highlighting the segmentation and blending of the word set. Then, the teacher taught them the segmentation and blending of the word set twice, so that the participants knew how to read out the word on their own by phonics rules. After the fourth training word was taught, the same practice was carried out one more time. Figure 3.3 below shows the teaching settings of the control group.
Figure3.3 Teaching Setting for the Control Group (Phonics Group) Data Analysis
The Statistical Package for the Social Sciences (SPSS) version 12.0 was utilized to perform quantitative analyses of data collected from participants’ decoding skills and changed attitudes in the pre-and post- tests. In answering research question one,
independent-samples t-tests were computed on the generalization post-test to measure the effectiveness of rime analogy and phonics training. For the second research question, given that the questions aimed to see attitudinal changes and participants’ perceived difficulties, data analysis was carried out in two ways: quantitatively for the questionnaire
26 At the beginning of the presentation and practice, the introduction of clue words was the same as for the experimental group.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
part and qualitatively in the interview part.
For the questionnaire part, English word reading attitudes were measured by scores of the questionnaires based on 4-point Likert scale. Independent-samples t-tests were first computed on the scores to detect if both groups were similar before training. Then, paired-samples t-tests were carried out for both groups separately, to see within-group
For the questionnaire part, English word reading attitudes were measured by scores of the questionnaires based on 4-point Likert scale. Independent-samples t-tests were first computed on the scores to detect if both groups were similar before training. Then, paired-samples t-tests were carried out for both groups separately, to see within-group