國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Sixteen weeks later Sixteen weeks later
Figure 3.3 The Whole Procedure of the StudyMain Study
Pilot Study
To verify the practicability of this study, the researcher carried out a pilot study, in which thirty-four students from a ninth-grade class who did not join the main study
A pilot study
A general proficiency test (EBCT)
A word selection test
A pre-test in word recognition
Experimental group:
The sound symbolism instruction
Control group:
The traditional translation-based approach
A post-test in word recognition +
A think-aloud task
A post-test in word recognition +
A think-aloud task
Interviews
Data analysis
‧
at the same school were chosen as the participants. The instruments included a vocabulary pre- and post-test (See Appendix G), the modified think-aloud protocols sheet (See Appendix H), and a questionnaire (See Appendix I). Basically, the pilot study followed the procedure of the main study, except for the scale. The participants were taught five new words in a period and the pilot study was conducted for a week, five periods. Because of the short-term instruction, the lesson only covered letter A, B and C illustrated by sound symbolism and the size of the target vocabulary was limited to twenty-five words. The teaching material (See Appendix J) for the target words was designed on the basis of Mo’s research (2005).
The results of the pilot study are as follows. (1) Before the instruction of sound symbolism, the mean score of the pre-test was 16.59, and after the instruction, the mean score of the post-test was 59.29. (2) The results of the modified think-aloud method indicated that the more the participants used sound symbolism in vocabulary memorization, the more vocabulary items they would remember. (3) The results of the questionnaires (See Appendix K) showed that more than 90%16 of the participants agreed that sound symbolism instruction was helpful for their vocabulary
memorization.
The pilot study was used as a foundation for the development of the main study, and two modifications were made. First, a test for word selection in the main study was designed. The results of the think-aloud method showed that some of the target words in the pilot study were not unknown words for the participants because the target words were selected only by the researcher. Therefore, a word selection test was added. Second, the questionnaire in the pilot study was developed for a preliminary investigation of participants’ attitudes toward sound symbolism instruction. However,
16 As shown in Appendix K, the percentage of the participants who agreed and strongly agreed that SS is helpful for English vocabulary memorization was 60% and 31.43% respectively, constituting a total of 91.43% in terms of positive responses.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
in the main study, the researcher intended to know more about participants’ views on the instruction of sound symbolism in depth. Consequently, those questions were modified by adding questions and included in the interviews in the main study.
Main Study
At the beginning, EBCT was administered to ensure that the two groups were homogeneous in regards to general English proficiency. Then, after word selection, a pre-test was utilized to ensure that the two groups were homogeneous in terms of the performance on the target vocabulary. During the instruction, the experimental group was taught sound symbolism to help them memorize vocabulary, while the control group was taught the same content with a traditional translation-based approach. After a sixteen-week instruction, a post-test was conducted to compare the two groups’
performances on the target vocabulary. In addition, the two groups were required to engage in a think-aloud task added to the post-test to minimize the chance of random correct responses in the post-test. Finally, ten volunteer participants from the
experimental group were given individual interviews.
Data Analysis
The data analysis of this project included both quantitative and qualitative analysis. First, the researcher used quantitative methods to analyze the data collected from the test sheets with the statistical software SPSS; second, the interviews were recorded and transcribed for qualitative analysis. The study was intended to find the answers to the three research questions: (1) Do students who receive the instruction of sound symbolism perform better than those who are taught the traditional
translation-based approach on vocabulary memorization? (2) Is the instruction of sound symbolism effective for high and low proficiency learners respectively? (3)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
How do students think about the instruction of sound symbolism? How can the instruction be improved?
To answer Research Question 1, t-tests were conducted as follows.
(1) The mean scores of EBCT of the two groups were compared by independent samples t-test to ensure that initially the two groups had the same level of general English proficiency.
(2) The mean scores of the pre-tests of both the experimental and control groups were compared by independent samples t-test to ascertain that the two groups had homogeneous performance on the target vocabulary initially.
(3) The mean scores of the post-tests of the two groups were compared by independent samples t-test to investigate the effectiveness of the instruction of sound symbolism. In addition, the results of the modified think-aloud method were analyzed to ensure that the participants’ performances on the post-test were not random guesses.
To answer Research Question 2, the mean score of the pre-test and that of the post-test of the experimental high subgroup were compared by paired samples t-test to evaluate whether there was significant progress. The identical procedure was
conducted for the experimental low subgroup to know whether the subgroup made significant progress.
Finally, to answer Research Question 3, the results of the interviews were analyzed qualitatively to find out the participants’ views on the instruction of sound symbolism and what effect the instruction of sound symbolism had on the participants as well as how to improve the instruction. The researcher organized the interviewees’
responses into appropriate categories on the basis of literature. The responses to the first part of the interview questions concerning the ways to memorize new words were categorized by Schmitt’s (1997) taxonomy of vocabulary learning strategies. The
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
responses to the second part of the interview questions regarding the views on the instruction of sound symbolism were examined by sound symbolism theories. The results of the post-tests, the information offered from the modified think-aloud method and the data gathered from the interviews were crosschecked.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER FOUR RESULTS
This chapter reports the results of the quantitative and qualitative analysis of the data collected. The results comprise four sections. The first two sections report the results for Research Question 1 and 2 — Do students who receive the instruction of sound symbolism perform better than those who are taught the traditional
translation-based approach on vocabulary memorization? Is the instruction of sound symbolism effective for high and low proficiency learners respectively? The third section presents the results for Research Question 3 — How do students think about the instruction of sound symbolism? How can the instruction be improved? The final section summarizes all of the findings in the study.
Results of the Pre-test and Post-test
This section presents the results of the pre-test and those of the post-test. 2.5 points are given when a participant answers a question correctly. The full score is 100.
Table 4.1 displays the mean scores of the pre-test and the post-test for the
experimental and control groups. As the statistics showed, the experimental group scored 7.64 and 57.08 in the pre-test and the post-test respectively, while the control group scored 8.19 and 35.42.
‧
Table 4.1 Statistics of Participants’ Pre-Test & Post-Test Scores
Results of the t-tests are presented in Table 4.2 and Table 4.3. The pre-test scores of the experimental and control groups showed no difference (t=2.36, p=0.814>0.05).
By contrast, there was a statistically significant difference in the post-test scores (t=3.076, p=0.003<0.05), with the experimental group manifesting fewer errors in the word recognition than the control group. The participants in the experimental group performed significantly better than their peers in the control group on the post-test.
Clearly, the findings indicate that the instruction of sound symbolism has a positive effect on student performance in terms of word recognition.
Table 4.2 Independent Samples t-test on Participants’ Pre-Test
Group
t-test for Equality of Means
‧
Table 4.3 Independent Samples t-test on Participants’ Post-Test
* p<0.05, ** p<0.01
As indicated in Table 4.4, in the experimental group, the High group made progress of 64.06 points whereas the Low group made progress of 38.5 points. The results of the paired samples t-test revealed that both the High group and the Low group made significant progress (High group: t=14.818, p=0.000<0.05; Low group:
t=6.452, p=0.000<0.05). It appears that the instruction of sound symbolism results in substantial differences in performance in word recognition for both high proficiency and low proficiency learners.
Table 4.4 Paired Samples t-test for Progress in the Experimental Group
* p<0.05, ** p<0.01, ***p<0.001
Levene’s Test for Equality of Variances
t-test for Equality of Means
‧
Results of the Think-aloud Method
As mentioned in Chapter 3, there were 40 vocabulary items in the post-test, each of which can be explained by sound symbolism. The participants were taught how to write down their thinking process in the training session. Afterwards, they were required to fill in the blanks of the think-aloud sheet (See Appendix D) with either an appropriate explanation of sound symbolism or other means of vocabulary
memorization17. Performances of the High and Low groups in the post-test and
think-aloud task are presented in Table 4.5, Table 4.6 and Figure 4.1. In Tables 4.5 and 4.6, the following information is presented: (1) the number of correct responses in the post-test, (2) the number of correct responses in the think-aloud sheet, (3) the
proportion of correct responses based on sound symbolism to total correct responses in the post-test, and (4) each participant’s score earned in the post-test. Among the correct responses in the post-test, two categories were further identified by
crosschecking the results of the post-test and those of the think-aloud sheet: (1) correct responses based on sound symbolism and (2) correct responses based on other means of vocabulary memorization. The purpose of the crosscheck was to ensure that the scores of the post-test were gained by the application of sound symbolism instead of other means of vocabulary memorization. Take Student One (S1) in Table 4.5 for example, he got 100 points on the post-test, which means he answered 40 questions correctly. By checking his think-aloud sheet, the researcher found that he completed 33 vocabulary items with appropriate explanations of sound symbolism. Consequently, among the 40 correct responses in the post-test, the researcher could conclude that 33 correct responses were made by the application of sound symbolism, while the other seven were based on other means of vocabulary memorization. Regarding the
17 In the experimental group, the results of the responses to the think-aloud sheet included either a blank or an explanation of sound symbolism. There was no response referring to other means of vocabulary memorization.
‧
proportion of correct responses based on sound symbolism to total correct responses in the post-test, the percentage was 82.5% (33/40=82.5%).
Table 4.5 and Figure 4.1 show the High group’s performance in the post-test and the think-aloud task. The results are as follows. (1) The scores of the High group on the post-test range from 42.5 to 100 points, with the mean score of 75.16 (See Table 4.5). (2) In terms of the proportion of correct responses based on sound symbolism to total correct responses in the post-test18, for around 60% (N=10)19 of the participants in the High group, their percentage was more than 80% in Table 4.5. Around 30%
(N=4)20 of the participants revealed the percentage with the distribution between 70%
and 80% and the remaining 10% (N=2)21 of the participants showed the percentage with the distribution between 50% and 70% in Table 4.5. (3) Overall, the proportion of correct responses based on sound symbolism to total correct responses in the post-test was quite high (See Figure 4.1), indicating a large number of students in the High group used sound symbolism to memorize the meaning of the new words in the post-test.
On the other hand, in the Low group (See Table 4.6 and Figure 4.1), among the data collected from the twenty participants, one piece of data information was invalid (See S8 in Table 4.6) because all of this participant’s correct responses in the post-test were random correct responses, which were based on guessing22. For the valid
nineteen pieces of data information, the results are as follows. (1) The scores of the Low group on the post-test range from 2.5 to 87.5, with the mean score of 43.16 (See Table 4.6). (2) In terms of the proportion of correct responses based on sound
18 Each of the vocabulary items in the post-test or the think-aloud sheet can be explained by sound symbolism.
19 See S1, S3, S4, S8, S9, S11, S12, S13, S14 and S15 in Table 4.5.
20 See S6, S7, S10 and S16 in Table 4.5.
21 See S2 and S5 in Table 4.5.
22 The information was obtained from an individual interview with the participant, who answered 13 questions correctly in the post-test and handed in his think-aloud sheet in blank at the same time.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
symbolism to total correct responses in the post-test, for around 70% (N=13)23 of the participants in the Low group, their percentage was more than or equal to 80% in Table 4.6. Around 20% (N=4)24 of the participants revealed the percentage with the distribution between 50% and 80%. For the remaining 10% (N=2)25 of the
participants, their percentage was 0%. (3) As a whole, the proportion of correct responses based on sound symbolism to total correct responses in the post-test was quite high (See Figure 4.1), indicating a great number of students in the Low group applied sound symbolism to memorizing the meaning of the new words in the post-test as well.
23 See S1, S2, S3, S4, S6, S7, S9, S10, S12, S16, S18, S19 and S20 in Table 4.6.
24 See S11, S13, S15 and S17 in Table 4.6.
25 See S5 and S14 in Table 4.6.
‧
Table 4.5 Performance of the High Group in the Post-test & Think-aloud Task
Participant
Correct Responses in the Post-test
Correct Responses in Think-aloud Sheet
Proportion of SS of Correct Responses in
‧
Performance of the High Group in the Post-test & Think-aloud Task
Participant
Correct Responses in the Post-test
Correct Responses in Think-aloud Sheet
Proportion of SS of Correct Responses in
Note: * SS referred to correct responses based on sound symbolism, while OM in this Table indicated those based on other means of vocabulary memorization. The column of figures were derived from dividing the number of SS by the number of correct responses in the post-test and presented in percentage. Take S1 for example, 82.5%
was the result of dividing 33 by 40 and expressed in percentage. The higher the percentage in the column was, the more application of sound symbolism the participant made in the post-test.
‧
Table 4.6 Performance of the Low Group in the Post-test & Think-aloud Task
Participant
Correct Responses in the Post-test
Correct Responses in Think-aloud Sheet
Proportion of SS of Correct Responses in
‧
Performance of the Low Group in the Post-test & Think-aloud Task
Participant
Correct Responses in the Post-test
Correct Responses in Think-aloud Sheet
Proportion of SS of Correct Responses in
Note: 1. * SS referred to correct responses based on sound symbolism, while OM in this Table
‧
indicated those based on other means of vocabulary memorization. The column offigures were derived from dividing the number of SS by the number of correct responses in the post-test and presented in percentage. Take S1 for example, 94.11%
was the result of dividing 32 by 34 and expressed in percentage. The higher the percentage in the column was, the more application of sound symbolism the participant made in the post-test.
2. **: invalid data information
Figure 4.1 Proportion of Correct Responses Based on Sound Symbolism to Total Correct Responses in the Post-test
Generally speaking, the result of the crosscheck is consistent with the
participants’ performances on the post-test. As indicated in Figure 4.1 and Table 4.7,
N of High Group= 16;
N of Low Group=20
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
up to 87% of those who performed well26 on the post-test made relatively a large number of correct responses based on sound symbolism in the think-aloud task. In terms of the proportion of correct responses based on sound symbolism to total correct responses in the post-test, their percentage was more than 80%27.
26 In this current study, the researcher defined the students with good performances on the post-test as those who got 70 points and above. See S1, S3, S7, S8, S9, S11, S12, S13, S14 and S15 in Table 4.5 and S1, S6, S12, S15 and S18 in Table 4.6.
27 See S1, S3, S8, S9, S11, S12, S13, S14 and S15 in the High group and S1, S6, S12, and S18 in the Low group in Table 4.7.
‧
Table 4.7 The Use of Sound Symbolism of the Participants with Good Performance Participants with
Good Performance*
(N=15)
Score of the Post-test Proportion of SS of Correct Responses in the Post-test
S1 100 82.50%
Number of participants whose
percentage was more than 13 (87%**) or equal to 80%
Note: 1. *Participants with good performance on the post-test were those who got 70 points and above on the post-test.
2. **The number was derived from dividing 13 by 15 and presented in percentage. 15
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
was the number of the participants with good performance on the post-test.
As a whole, as indicated in Tables 4.5 and 4.6, for the experimental group, the more the participants used sound symbolism in vocabulary memorization, the more vocabulary items they would remember. In addition, the results of the think-aloud method showed that in general most of the correct responses in the post-test were based on vocabulary memorization to which sound symbolism was applied. Stated it in another way, in the experimental group, most of the scores of the post-test were gained by the use of sound symbolism.
However, the results of the think-aloud method did not mean that the use of sound symbolism led to no mistakes in the think-aloud task. Around 61% (N=22) of the participants made one to four inappropriate applications of sound symbolism in the thinking process, resulting in incorrect answers in the think-aloud sheet. By examining those mistakes, the researcher found that most of them included the same consonant clusters. For example, many participants confounded the meanings of the words pressure with primary because in the instruction of sound symbolism, both of them were explained by the sound symbolic information of the initial pr- sound, which is suggestive of not only the action of pressing but also the quality of being major or chief (See Appendix F, p.126-127).
Meanwhile, the results of the think-aloud task in the control group showed that all of the correct responses in the post-test were on the basis of either vocabulary memorization by rote or random correct responses, which means correct responses given by chance.
Results of the Interviews
After the post-test and the think-aloud task, ten volunteer participants from the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
experimental group, including five high proficiency students and five low proficiency students, were interviewed individually. The researcher distributed sheets of interview questions (See Appendix E) to the interviewees right before each individual oral interview and gave each of them fifteen minutes to prepare so that the interviewees had time to think about how to answer the interview questions. During the preparation, students were not allowed to discuss with each other. Due to the limited English
proficiency of the ninth graders in junior high school, these interviews were
conducted in Chinese (See Appendix L). Each interview lasted for approximately ten minutes.
conducted in Chinese (See Appendix L). Each interview lasted for approximately ten minutes.