3. METHOD
3.4 Procedures of the Formal Study
3.4.3 Data Analysis
After the calculation of cohesive devices in high and low proficiency writers’
samples, both quantitative and qualitative analyses were indispensable for answering the four research questions. With regard to the quantitative analysis of cohesive devices, the statistical tool, Independent-Samples t-test, was applied and the P value was set at .05 probability to examine whether there was significant difference between groups or genres. After all the writing samples were collected, the number of cohesive devices, T-units11 in each sample, total number of cohesive ties in each category of cohesive devices, and percentage of the cohesive device in each category were calculated and presented in tables. To answer the first two research questions, the
11 T-unit is first introduced by Kellogg Hunt (1965) to measure development of sentences in the writing of grade-school children. A T-unit-based analysis is suitable for low-level compositions where run-on sentences are common. Each T-unit consists of one main clause and its dependent clauses.
52
writer’s proficiency was regarded as a variable. Therefore, high and low proficiency writer’s use of cohesive devices were compared first in narration and then in exposition. Next, by holding the factor, proficiency level, constant, we examined whether high or low proficiency writers’ use of cohesive device in narrative and expository writing would be different.
Apart from observing the different uses of cohesive devices quantitatively, a qualitative analysis would be carried out as well to further examine whether there was any inappropriate use of cohesive devices or errors and how students used cohesive ties to cohere their written discourse. The inappropriate use of cohesive devices included cohesive devices that had the extra addition or omission of the essential cohesive devices, wrong choices of words, wrong inflection, improper rhetoric and so on. First, inappropriate use of cohesive devices in high and low proficiency writer’s narrative writing would be examined, followed by those in both groups’ expository writing. Next, high proficiency writers’ errors in narration and exposition, and low proficiency writers’ errors in two genres would be scrutinized respectively. The purpose of this was to see whether there were patterns of errors corresponding to particular genres. In addition to the calculation of numbers of inappropriate cohesive devices, writers’ errors would be enumerated and analyzed from the discourse perspective. Also, correction of the students’ writing would also be listed.
53
CHAPTER FOUR RESULTS AND DISCUSSION
This chapter reports and discusses the results of the study. The statistical analysis used in this study was t-test. The whole chapter is organized into two parts.
The first part is the quantitative analysis of participants’ cohesive devices. The data were collected from senior English majors and freshmen non-English majors. And the statistical results of each cohesive device will be presented and discussed. The second part is the qualitative analysis of participant’s inappropriate use of cohesive devices.
That is, high and low proficiency groups’ inappropriate use of cohesive devices in narrative writing will be presented, followed by high and low proficiency groups’
inappropriate use of cohesive devices in expository writing. Finally, we will examine both groups’ inappropriate use of cohesive devices in the two genres.
4.1 Quantitative Analysis
In the section, the quantitative analysis of cohesive devices in terms of high and low proficiency groups and different genres, narration and exposition, will be presented and discussed.
4.1.1 High and Low Proficiency Groups in Narrative Writing
In response to the first research question, high and low proficiency level students’ total number of cohesive devices in narrative writing were tallied and compared. Table 1 presents the number of cohesive devices across proficiency levels and across modes. High proficiency writers used a total of 902 cohesive devices and low proficiency writers used a total of 615 cohesive devices when writing in the narrative mode. To answer the first research question, a t-test reveals that there was a significant difference in high and low proficiency level students’ use of cohesive devices in narrative writing at the .05 level of probability, as indicated in Table 2;
namely, the total number of cohesive devices in the high proficiency group was
54
significantly greater than that of the low proficiency group in the narrative mode.
Table 1
High and Low Groups’ Use of Cohesive Devices in Narration and Exposition
Narration Exposition
Comparison of Total Number of Cohesive Devices in Narration between Two Groups Level N Mean Std. Deviation Std. Error Mean T-value
4.1.1.1 High and Low Proficiency Groups’ Use of Grammatical Cohesive Devices A further t-test was conducted to find out which type of cohesive devices, grammatical or lexical, produced a significant effect in high and low proficiency level writers’ narrative writing, as shown in Table 3. As can be seen from the table, there was a significant difference in the number of grammar cohesive devices employed in high and low proficiency groups’ narrative writing. That is, high proficiency writers employed a greater number of grammatical cohesive devices in narrative writing;
whereas there was no significant difference in the number of lexical cohesive devices in high and low proficiency groups’ narrative writing. In other words, there was no difference in the number of lexical cohesive devices used by both groups in narrative writing.
55
Table 3
Grammatical and Lexical Cohesive Devices in High and Low Groups in Narration H/L N Mean Std. Deviation Std. Error Mean T-value
Note. Gram= Grammatical cohesive devices; Lexicon= Lexical Cohesive devices.
*p < .05.
Since high proficiency writers had a significantly greater number of grammatical cohesive devices than low proficiency writers, a further exploration of the subcategories of grammatical cohesive devices between the two groups in narrative writing was carried out to determine which subcategory of the grammatical cohesive devices in the high proficiency group significantly outnumbered the subcategory of the grammatical cohesive devices in the low proficiency group in narrative writing.
First, a complete list of the four subcategories of grammatical cohesive devices was presented in Table 4. The table shows that both reference and conjunction had comparatively greater frequency of occurrence in the two groups. In addition, in both groups, reference had the highest frequency of occurrence, followed by conjunction, substitution and ellipsis in a descending order. The comparison of the number of reference cohesive devices between the two groups was made by a t-test. As shown in Table 5, there was a significant difference in high and low groups’ use of reference cohesive devices in narration. Second, the number of substitution was compared between high and low proficiency groups. A t-test reveals that there was no significant difference between the two groups, as indicated in Table 6. Third, because no instances of ellipsis were found between two groups, there was no need to make the comparison. Fourth, the number of conjunction between the two groups was investigated in narration. A t-test reveals that there was a significant difference in the
56
use of conjunction as cohesive devices between high and low proficiency groups in narrative writing, as indicated in Table 7. Furthermore, of the sub-forms of reference, pronominal reference and definite article were proved to show significant difference between the two groups, as shown in Tables 8-10 (See Appendix H, p. 140). Also, of all the four sub-forms of conjunction, additive, adversative, causal and temporal conjunction, high proficiency group’s use of cohesive devices in additive and temporal conjunction was significantly different from those in the low proficiency group’s, as indicated in Tables 11-15 (See Appendix H, p. 140-141).
Table 4
Distribution of Grammatical Cohesive Devices in Narration
Grammatical Cohesion High Low TOTAL
(H+L)
Comparison of Numbers of Reference in the Two Groups in Narration
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Substitution in the Two Groups in Narration
Level N Mean Std. Deviation Std. Error Mean T-value
57
Table 7
Comparison of Numbers of Conjunction in the Two Groups in Narration
Level N Mean Std. Deviation Std. Error Mean T-value
4.1.1.2 High and Low Proficiency Groups’ Use of Lexical Cohesive Devices With regard to the distribution of lexical cohesive devices between the two groups in narration, the total number of lexical cohesive devices in high and low proficiency groups is shown in Table 16. In the high proficiency group, repetition has the highest frequency, followed by synonym, superordinate, and antonym while in the low proficiency group, repetition has the highest frequency, followed by superordinate, synonym and antonym. Moreover, there was a significant difference in the use of synonym and antonym between the two groups by the application of t-tests, as shown in Tables 18 and 19. The result reveals that in narration, high proficiency writers had a wider vocabulary base and were more capable of displaying their skills of paraphrasing by using various words or phrases to express the synonymous or antonymous relationship.
Table 16
Distribution of Lexical Cohesive Devices in Narration
Lexical Cohesion High Low TOTAL
(H+L)
58
Table 17
Comparison of Numbers of Repetition in the Two Groups in Narration
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Synonym in the Two Groups in Narration
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Antonym in the Two Groups in Narration
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Superordinate in the Two Groups in Narration
Level N Mean Std. Deviation Std. Error Mean T-value
4.1.2 High and Low Proficiency Groups in Expository Writing
To address the second research question, that is, whether there were any significant differences in high and low proficiency level students’ use of cohesive devices in exposition, high and low proficiency level students’ cohesive devices were computed through a t-test. High proficiency level students used 673 cohesive devices and low proficiency level students, 504 cohesive devices in exposition, as shown in
59
(Table 1, p. 54). As presented in Table 21, there was a significant difference in high and low proficiency level students’ use of cohesive devices in exposition at the .05 level of probability. Further, high and low proficiency level students’ grammatical and cohesive devices in the expository writing were compared. The results show that there was a significant difference in both high and low proficiency groups’ use of grammatical and lexical cohesive devices in exposition, as shown in Table 22.
Table 21
Comparison of Total Number of Cohesive Devices in Exposition between Two Groups Level N Mean Std. Deviation SStd. Error Mean T-value
Grammatical and Lexical Cohesive Devices in High and Low Groups in Exposition H/L N Mean Std. Deviation Std. Error Mean T-value
Note. Gram= Grammatical cohesive devices; Lexicon= Lexical Cohesive devices.
*p < .05.
4.1.2.1 High and Low Proficiency Group’s Use of Grammatical Cohesive Devices To explore the distribution of the subcategory of grammatical cohesive devices, reference, substitution, ellipsis and conjunction, a complete list of their occurrences between the two groups is shown in Table 23. Both reference and conjunction show the greater frequency of occurrence in the two groups. Besides, in both groups, reference has the highest frequency of occurrence, followed by conjunction, substitution and ellipsis in a descending order. Despite the higher frequency of occurrence in the high proficiency group, the comparison of the reference,
60
substitution, ellipsis and conjunction between the two groups in exposition was not significant, as indicated in Tables 24-27. Moreover, of all four sub-forms of conjunction, additive, adversative, causal and temporal conjunction, there was no significant difference in high and low proficiency level writers’ use of additive, adversative, causal and temporal conjunction in exposition, as indicated in Tables 28-32 (See Appendix H, p. 142-143).
Table 23
Distribution of Grammatical Cohesive Devices in Exposition
Grammatical Cohesion High Low TOTAL
(H+L)
Comparison of Numbers of Reference in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Substitution in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Ellipsis in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value
61
Table 27
Comparison of Numbers of Conjunctions in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value
4.1.2.2 High and Low Proficiency Group’s Use of Lexical Cohesive Devices With regard to the distribution of lexical cohesive devices between high and low proficiency groups in exposition, the total number of lexical cohesive devices in high and low proficiency groups is shown in Table 33. In the high proficiency group, repetition has the highest frequency, followed by synonym, antonym and superordinate while in the low proficiency group, repetition has the highest frequency, followed by antonym, synonym and superordinate (Table 33). Moreover, there was a significant difference in the use of synonym and superordinate between the two groups by the application of t-tests, as shown in Tables 35 and 37. The results reveal that in exposition, high proficiency writers not only were more capable of displaying the skills of paraphrasing by using greater variety of words to express the synonymous relationship but also were more capable of using superordinate to present the general ideas first, followed by detailed examples to strengthen their points of views than low proficiency writers did in exposition.
Table 33
Distribution of Lexical Cohesive Devices in Exposition
Lexical Cohesion High Low TOTAL
(H+L)
62
Table 34
Comparison of Numbers of Repetition in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Synonym in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Antonym in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value
Comparison of Numbers of Superordiante in the Two Groups in Exposition
Level N Mean Std. Deviation Std. Error Mean T-value both narration and exposition, respectively. The total number of cohesive devices used by the high proficiency group outnumbered that used by the low proficiency group, which seemed to indicate that both in narration and exposition, the proficiency level influenced the employment of cohesive devices. That is, high proficiency group
63
writers were not only better at elaborating their ideas but also better at using greater variety of grammatical or lexical cohesive devices to make their writing more condensed and coherent. The findings also confirm previous studies suggesting that the total number of cohesive devices was correlated positively with the quality of writing and the number of cohesive devices may serve as a tool to distinguish between stages of writing development (Ferris, 1994; Liu & Braine, 2005; McCulley, 1985;
Normant, 1982; Witte & Faigley, 1981).
When comparing the grammatical cohesive devices used by high and low proficiency groups, there was a significant difference in the employment of the grammatical cohesive devices between the high and low proficiency groups in writing the narrative essays. This was also found in high and low proficiency groups’ use of the grammatical cohesive devices in expository writing. Also, when we compared the use of reference, substitution, ellipsis and conjunction in high and low proficiency writers’ writing, in narration, reference and conjunction was found to differ significantly from the other two types while in exposition, none of them revealed any significant differences, although the mean and p value of the conjunction almost reached significance. Further, in the narrative writing, high proficiency level writers not only significantly used more pronominal reference and definite articles but also more additive and temporal conjunction than low proficiency level writers did. This seemed to suggest that high proficiency writers were not only better at elaborating the stories by establishing stronger referential cohesive bonds but also had clearer concept of the temporal progression in the narrative stories, so that they were more skillful in applying those reference and conjunctive cohesive elements to make the discourse flow coherent. This finding also is similar to that in Jin’s (2001) study that high proficiency writers used significantly more personal reference and temporal conjunction.
64
On the other hand, observing the frequency and percentage of reference, substitution, ellipsis and conjunction in high and low proficiency groups’ writing, it was found that in narration and in both groups, reference was the dominant cohesive devices, followed by conjunction, substitution and ellipsis. This same pattern occurred in the high and low proficiency writer’s expository writing as well. This phenomenon is consistent with previous studies (Chang, 1995; Liang, 1997; Liu & Braine, 2005;
Norment, 1982) and also confirms Fine’s claim that substitution or ellipsis often takes up less than 3 % of all the grammatical cohesion devices, and thus these two are called “minor forms of cohesion” (Fine, 1994, p. 8).
With regard to the comparison of lexical cohesive devices, a significant difference was found in exposition between high and low proficiency groups, which may suggest that the exposition may be a more difficult genre in which for low proficiency writers to fully elaborate their arguments or viewpoints by applying words with various semantic relationships. Besides, both high and low proficiency level writers used repetition most in their writing. The results are similar to those of previous findings (Chang, 1995; Chen, 2003; Liang, 1997) and probably indicate that repetition of the key idea is essential when elaborating the topic and concept and that it is also easier for writers of different proficiency levels to employ in writing.
Additionally, in narration, there were significant differences in the employment of synonym and antonym while in exposition, there was a significant difference in the employment of synonym and superordinate. From the comparison of different proficiency writers, it is clear that high proficiency writers differed from low proficiency writers in terms of their ability to manipulate greater diversity of synonymous words whether in narration or exposition. The results confirm the previous findings that the greater use of synonyms correlated positively with better writing (Chen, 2003; Crowhurst, 1987; Jin, 2001).
65
4.1.4 High Proficiency Group in Narrative and Expository Writing
The third research question is to explore whether there is a significant difference in the high proficiency level students’ use of cohesive devices when writing in the two modes, narration and exposition. First of all, the total number of cohesive devices cohesive devices between the two genres reached the point of significant difference.
The results show that there was a significant difference in the high proficiency group’s use of grammatical cohesive devices between narration and exposition while no significant difference was found in the high proficiency group’s use of lexical cohesive devices, as shown in Table 40.
Table 38
High Proficiency Group’s Cohesive Devices in Narrative and Expository Writing High Proficiency Group
High Proficiency Group’s Cohesive Devices in the Two Genres
Genre N Mean Std. Deviation Std. Error Mean T-value Narration 9 100.2222 36.63256 12.21085
Exposition 9 74.7778 17.45550 5.81850
T=1.881 P=.078 Note. *p < .05.
66 Note. Gram= Grammatical cohesive devices; Lexicon= Lexical Cohesive devices.
*p < .05.
Third, to examine whether there was difference in the subcategories of grammatical cohesive devices in the two genres, a complete frequency count of each subcategory was conducted, the result of which is shown in Table 41. As can be seen from the table, in the two genres reference has the largest proportion of frequency, followed by conjunction, substitution, and ellipsis. A t-test was then conducted and subsequently revealed that there was a significant difference in the high proficiency
Third, to examine whether there was difference in the subcategories of grammatical cohesive devices in the two genres, a complete frequency count of each subcategory was conducted, the result of which is shown in Table 41. As can be seen from the table, in the two genres reference has the largest proportion of frequency, followed by conjunction, substitution, and ellipsis. A t-test was then conducted and subsequently revealed that there was a significant difference in the high proficiency