3.4 The Analysis
3.4.3 Evaluation of Thematic Choices (TCs)…
The current research analyzed participants in unmarked themes, and logical connectors in marked themes with the purpose of investigating whether coherence was achieved at the sentence level and answering the second research question. To examine the occurrence of each participant type and the variety of participants, the researcher tallied the number of each type. Each occurrence of the participant type
76
was then divided by the total number of participants to calculate the percentage of
each participant type both in pretest and posttest writings.
For instance, “Cindy” was the unmarked theme of the first sentence in Table 3.3,
“Cindy was invited to her best friend’s birthday party few days ago.” It was also the
participant of the sentence, belonging to the people participant type. There were
seven themes and all of them were classified as people participant type, which dominated unskillful student writers’ essays (Hawes & Thomas, 1997; Liu & Liu,
2013; Mellos, 2011). It is suggested that the writer vary the participant type. The study then explored whether there was any difference in the subjects’ use of
participant types after the intervention.
Next, occurrences of logical connectors in marked themes were counted and the
percentage of each logical connector was calculated. Adapted from Hyland (2004a), five types of logical connectors, which frequently appeared in the learners’ picture
writings, were categorized. They were sequential logical connectors, the logical connectors that developed cause and effect sets of ideas, the ones that compared and contrasted, the ones that provided explanation, and the ones that summarized the event.
For instance, in Table 3.3, “At the same time” was the sequential logical connector employed in the marked theme to indicate the time and the sequence of
77
events. Nevertheless, it was the only type of logical connector in the paragraph. Again, more marked theme types were recommended to be included in student writings. The aims of the analysis were to investigate the variety of logical connectors both before
and after the intervention and to uncover whether there was any difference in the students’ use of logical connectors in marked themes after the intervention.
In sum, the quantitative analysis might help reveal whether the participants’
overall writing scores improved after the intervention. The qualitative analysis consisted of three parts, i.e. the examination of rhetorical moves, TPs and TCs in picture writings, to unveil the efficacy of SFL genre-based instruction conducted in EFL context.
78
79
CHAPTER FOUR
Results
The goal of the current study was to investigate how the instruction informed by SFL influenced EFL senior high school students’ picture writings. Both quantitative and qualitative approaches were adopted to analyze the pretest and posttest writing scores, rhetorical structures, TPs and TCs in the pretest and posttest data. The results were organized in two sections: (1) the pretest and posttest scores of learner writings, (2) the learners’ performances in TPs, rhetorical structures, and TCs, with a focus on participants in the unmarked themes and logical connectors in the marked themes.
4.1 The Pretest and Posttest Picture Writings Scores
A quantitative approach was first adopted to answer the first research question.
Thirty students received the instruction of TCs, TPs, and rhetorical structures. The students then completed the pretest and posttest picture writing tasks before and after the three-week instruction respectively. On the pretest, the total number of the words used in the thirty pretest writings was 4876 words. On average, there were 162.53
80
words in one piece of pretest writing. An increase of words was observed on the posttest. 4951 words were found in the thirty posttest writings. The average number of the words used in one piece of writing rose to 165.03 words after the instruction informed by SFL. Regarding the number of sentences in the learner writings, 415 sentences were identified in the pretest and, on average, the students wrote 13.83 sentences in one piece of pretest writing. The learners were also observed to write
more sentences after the instruction. There were 451 sentences in total and, on average, 15.03 sentences were identified in one piece of posttest writing.
Next, the learner writings were rated according to GSAT scoring rubrics and scores were given to the five criteria, content, organization, grammar, word choice, and mechanics. Based on the five criteria, a total score was given for the overall quality of each student writing. Eight out of the thirty writings were graded by the second rater and the inter-rater reliability was calculated using the Pearson correlation coefficient. The analysis revealed a high level of agreement between the two raters (Pearson r=0.951). Next, a paired t-test was conducted to compare the students’
overall performances and scores on the five constructs, i.e. content, organization, grammar, word choice, and mechanics, before and after the intervention. The results of the paired t-test are reported in Table 4.1.
81
Table 4.1 The Results of Paired-T Test on the Pretest and Posttest Picture
Writing Tasks Scores
Constructs Mean Standard Deviation df T-value P-value
Pretest Posttest Pretest Posttest
Content 2.350 2.867 0.709 0.490 30 -3.59 0.001*
Organization 2.317 3.233 0.725 0.487 30 -6.11 0.000*
Grammar 1.8333 1.7500 0.4795 0.3655 30 0.92 0.366 Word Choice 1.6500 1.8833 0.4385 0.4676 30 -2.96 0.006*
Mechanics 1.3500 1.5500 0.3256 0.3560 30 -2.47 0.020 Overall 9.500 11.283 1.925 1.350 30 -5.37 0.000*
* p<0.01
As Table 4.1 presented, the scores for pretest (M = 9.5, SD = 1.925) and posttest writings (M = 11.283, SD = 1.35) differed significantly (t = -5.37, p = 0.000). The statistics indicated that the score gains after the instruction reached a significance level of p<0.01. The participants improved by 1.783 points after the intervention and the p-value of the overall score was 0.000. The findings were in line with earlier studies (Brown & Marshall, 2012; Chen & Su, 2012; Cheng, 2008; Henry &
Roseberry, 1998; Mellos, 2011), which also observed significant improvement after
82
the instruction of TCs, TPs and rhetorical structures. The results implied that EFL learners benefited from the instruction informed by SFL.
Furthermore, a paired t-test was performed to investigate whether, after the intervention, there was significant difference on each of the five constructs, i.e.
content, organization, grammar, word choice, and mechanics. Significant differences
were found between the pretest and posttest scores of content, organization and word choice. As summarized in Table 4.1, the mean difference of the content scores was 0.517 (t = -3.59, p = 0.001), which reached significance. The participants made even larger improvement on the organization scores. The mean difference was 0.917 (t = -6.11, p = 0.000), which was also statistically significant. The results supported previous studies that the instruction informed by SFL improved the content and organization scores and also helped EFL learners produce coherent texts with rich
content (Brown & Marshall, 2012; Cheng, 2008; Henry & Roseberry, 1998).
Next, significant difference of learners’ word choice scores was also found with the mean scores rising by 0.2333 (t = -2.96, p = 0.008). The instruction program dealt
with the frequent participants and logical connectors in picture writings. The instruction of the meaningful units together with the hands-on practices might thus contribute to the significant increase of word choice. It was a supplementary benefit of the instruction.
83
On the other hand, no significance was observed in the participants’ grammar scores (t = 0.92, p = 0.366). Similarly, the increase of the scores on mechanics was 0.2, which did not reach significance (t = -2.47, p = 0.02). A possible explanation for the findings might be that the instruction informed by SFL focused on the discourse structure of picture writings and introduced the linguistic features associated with the genre, including the TCs and TPs, to improve coherence. Thus, it was not surprising that the differences of scores on the two constructs, i.e. grammar and mechanics, were not significant. The results corroborated previous research that the progress in content and organization was larger than that in word choice and grammar (Chen & Su,
2012).
4.2 Coherence in the Learner Writings
The students’ improvement in coherence could be further verified by examining
the TPs, TCs, at the sentence level and rhetorical moves at the discourse level. The
succeeding section attempted to answer the second research question which examined
whether coherence improved in the following three aspects, i.e. TCs, TPs and rhetorical moves, in the posttest.
4.2.1 TC in the Learner Writings
84
The current study examined how the instruction informed by SFL influenced EFL learners’ picture writings. TCs in the EFL senior high school learner writings were the first research foci in the current study. Specific emphasis was laid on participants in unmarked themes and logical connectors in marked themes. The
following sub-sections elaborated on the findings of the TC analysis.
Evaluation of Participants in Unmarked Themes The present study investigated
the variety of participant types in the pretest and posttest picture writings. The participant types identified in all the pretest and posttest writings were people, object,
dummy, intangible entity, and referent. The frequency and percentage of each type were calculated to judge if coherence in the aspects of TCs improved. The results obtained from the participant analysis of the thirty pretest and thirty posttest writing were summarized in Table 4.2.
Table 4.2 Participants in Unmarked Themes of the EFL Senior High School
Learners’ Pretest and Posttest Picture Writings
Pretest Posttest
Types Frequency Percentage Frequency Percentage
People 327 81 % 336 80 %
85
Object 20 5 % 35 8 %
Dummy 31 8 % 19 4 %
Intangible
entity
23 6 % 32 8 %
Referent 1 0 % 0 0 %
Total 402 100 % 422 100 %
Note. Percentage = frequency/ the total number of participants
It can be seen in Table 4.2 that, both in the pretest and posttest picture writings, the students used people participants most of the time. There were 327 cases in the pretest and 336 cases in the posttest. It accounted for 81% and 80% in the pretest and posttest respectively. The result was in accordance with the previous studies (Cheng, 2008; Correa & Domínguez, 2014; Derewianka, 2004), which reported that people participant was the dominant type in narratives. Figure 4.1 displayed the percentage
of each participant type in the pretest writings on the left and the percentage of each participant type in the posttest writings on the right.
86
Figure 4.1 Participants in Unmarked Themes of the Pretest and Posttest Picture
Writings
Closer inspection of Figure 4.1 showed that the people participants were the most prevalent among other participant types. Next, the dummy participant, such as
“it” and “there”, was also frequently found in the pretest writings. 31 cases were identified and accounted for 8% on the pretest. Participants belonging to the object and intangible entity type appeared more than 20 times, accounting for 5% and 6%
respectively in pretest and posttest student writings. However, the referent type, such as “that”, was only used once.
In the posttest, the major type was still people participant. 336 cases were observed and the percentage was 80%. The frequency rose, but the percentage slightly decreased because the use of other participant types also increased. It was encouraging in that the students diversified the participant types, which added variety to the posttest picture writings. Another evidence of the positive influence was that, in
People
87
addition to the people participant, the number of object and intangible entity grew to more than 30 cases and the percentage also rose to 8%. The number and frequency of dummy declined and there was no use of the referent type.
Overall, the students employed a large number of people participants and object participants on both tests, which accounted for 86% and 88% in total. The findings
corroborated the research literature, which indicated that EFL learners often used content nouns, or simple nominal groups in various genres (Hawes & Thomas, 1997;
Liu & Liu, 2013; Mellos, 2011). Usually, picture writings were descriptions of characters, so the students mostly relied on people and object participants. The learner writings could be improved if the students varied the participant types by describing from different perspectives or using a wide variety of sentence structures, such as narrating from the passive voice.
Evaluation of Logical Connectors in Marked Themes Adapted from Hyland
(2004a), the logical connectors identified in the pretest and posttest writings were categorized into five types, i.e. the ones that compared and contrasted, explained the sequence of events, providing explanation, showed the cause and effect of events, and offering conclusion. To determine if the learners’ use of logical connectors in marked themes differed in the pretest and posttest, the frequency and percentage of each type
88
were calculated. The overall findings are provided in Table 4.3.
Table 4.3 Logical Connectors in Marked Themes of the EFL Senior High School
Learners’ Pretest and Posttest Picture Writings
Pretest Posttest
Types Frequency Percentage Frequency Percentage
Contrast 9 18 % 9 14 %
Sequence 23 47 % 30 46 %
Explanation 5 10 % 3 4 %
Cause and
Effect
11 23 % 20 31 %
Conclusion 1 2 % 3 5 %
Total 49 100 % 65 100 %
Note. Percentage = frequency/ the total number of logical connectors in unmarked
themes
Table 4.3 showed that the use of logical connectors increased after the instruction.
The number of logical connectors rose from 49 cases in the pretest to 65 cases in the posttest. 15 students, i.e. half of them, employed more logical connectors in posttest
89
writings, while 7 students, i.e. 23% of them, employed the same number of logical connectors in marked theme positions. According to the literature (Cheng, 2008;
Hawes & Thomas, 1997; Mellos, 2011; Vande Kopple, 1991), logical connectors improved coherence. The increase of logical connectors might contribute to the progress of students’ overall performance. Figure 4.2 presented the percentage of each logical connector type in the pretest writings on the left and the percentage of each logical connector type in the posttest writings on the right.
Figure 4.2 Logical Connectors in Marked Themes of Pretest and Posttest Picture
Writings
As the above figures showed, both in the pretest and posttest, the most widely used type was to describe a sequence, such as “then”, “next” and “first”, which accounted for 47% and 46% in the pretest and posttest writings respectively. It was
Contrast,
90
quite common for learners to use sequential logical connectors to tell stories and describe events when composing picture writings. For instance, the students usually used “then” to narrate a series of events, as in “I ran as fast as I could to leave there and immediately threw that coat away. Then, I bought the same windbreaker and went to a park to take a break.” The encouraging findings were in line with Albufalasa (2013), who also reported that the use of sequential logical connectors increased after the instruction informed by SFL.
Second, there were also many logical connectors that developed cause and effect sets of ideas, including “therefore” and “as a result”, accounting for 23% on the pretest and 31% in the posttest. Frequency of the type went from 11 to 20 after the instruction, which almost doubled after the intervention. The nature of the genre, i.e.
picture writings, might explain why this type of logical connectors frequently appeared. To complete picture writing tasks, the learners were required to tell stories and wrote about a series of cases. Therefore, the learners often used such logical connectors to explain why the event took place and what led to the outcome of the story.
The next type commonly found in the student writings was the logical connectors that compared and contrasted, like “however”. Nine cases were found and the percentage was 18% in the pretest and 14% in the posttest. Fourth, the logical
91
connectors that provided explanation or added ideas, such as “besides” and “in fact”, were fewer. The learners employed this type of logical connectors five times, accounting for 10%, in pretest writings and three times, accounting for 4%, in posttest writings. Finally, the logical connectors that concluded or summarized the event, including “to sum up”, just occurred once, accounting for 2%, in the pretest and three times, 5%, in the posttest. The low frequency of the three types, i.e. logical connectors that compared and contrasted ideas, added different ideas, and offered conclusions, were not surprising because the learners were to tell stories in picture writings. Thus, it would be rare to find the logical connectors that compared, contrasted, and summarized ideas in picture writings.
In summary, the learners employed more logical connectors to connect sentences after the instruction informed by SFL. For instance, sequential logical connectors and logical connectors that developed cause and effect set of ideas were adopted to narrate the events of picture writings. Furthermore, various types of logical connectors were also used to improve coherence in the learners’ posttest writings. After the instruction, coherence in the learners’ picture writings improved with the number and types of logical connectors increasing in the learners’ picture writings.
4.2.2 TP in the Learner Writings
92
As mentioned in Chapter 2, TPs in the learner writings helped improve coherence at the sentence level and the instruction of TPs might help students produce coherent texts (Halliday, 1985). Thus, to answer the second research question, the TP patterns in the student writings were also analyzed after the instruction of TPs. The three TP types identified in the pretest and posttest writings were constant TP, linear TP and derived hyper TP. TP analysis was conducted to verify the efficacy of the intervention. The overall findings are presented in Table 4.4.
Table 4.4 Constant TP, Linear TP, and Derived Hyper TP in the EFL Senior High
School Learners’ Pretest and Posttest Picture Writings
Pretest Posttest
Types Frequency Percentage Frequency Percentage
Constant 189 94 % 122 79%
Linear 6 3 % 27 17 %
Derived
Hyper
6 3 % 6 4 %
Total 201 100 % 155 100 %
Note. Percentage = frequency/ the total number of TPs
93
As shown in Table 4.4, the use of constant TP was much more than any other TP type both in the pretest and posttest. 189 cases were identified in the pretest and 122 cases in the posttest, which accounted for 94% and 79% respectively. The learners mainly relied on constant TP in picture writings. The results corroborated previous research, which suggested that constant TP characterized EFL learner writings in various genres (Christie & Dreyfus, 2007; Hawes & Thomas, 1997; Liu & Liu, 2013;
Mellos, 2011; Wang, 2007). In picture writings, the story usually revolved around the main characters; as a result, the learners described what happened to the main characters and employed the same people or object participants as the theme.
Besides, even though more sentences were used, the total number of TPs decreased from 201 to 155 cases. Among all the TP types, the number of constant TPs dramatically dropped from 189 to 122. The reason might be that the learners avoided using constant TPs and varied the participant type in the posttest. Because the participants in the theme positions were different, constant TPs were less likely to be
formed. Yet, many learners were unable to implement other TP types, so the total number of TPs decreased. Figure 4.3 presented the percentage that each TP type accounted for in the pretest writings on the left and the percentage of each TP type in the posttest writings on the right.
94
Figure 4.3 TPs in the Pretest and Posttest Writings
Figure 4.3 demonstrated that constant TP dominated both the pretest and posttest writings. First, linear TP was identified six cases in the pretest writings and 27 cases in posttest writings. The number skyrocketed after the TP instruction. The percentage of linear TP rose from 3% to 17%, which almost tripled in the posttest. The results supported previous research that linear TP, an indicator of coherent writings, appeared more often after the instruction of TPs (Albufalasa, 2013; Liu & Liu, 2013; Mellos, 2011).
Finally, derived hyper TP appeared six times both in the pretest and posttest. The percentage slightly grew from 3% to 4% as the number of constant TP reduced. It was observed that the number and percentage of constant TP decreased in the posttest and the other TP types grew in number. The students tended to vary their TP types after
Finally, derived hyper TP appeared six times both in the pretest and posttest. The percentage slightly grew from 3% to 4% as the number of constant TP reduced. It was observed that the number and percentage of constant TP decreased in the posttest and the other TP types grew in number. The students tended to vary their TP types after