Studies of Writing across Different Genres

When it comes to studies of the cohesion, coherence and writing quality, it appears that the above-mentioned studies limited their focus to the performance of cohesion across different proficiency groups. Indeed, research on cohesion comparing two or three different genres was rare. And few attempted to link the variety of genres with writers of different proficiency levels when examining what distinguishes good from poor writing in the use of cohesion. However, there were still studies that attempted to investigate whether writers at different stages of development would perform distinctively in their employment of cohesive devices or whether difference in genre influences the distribution of cohesive devices.

In 1982, Norment conducted a study to compare the differences and similarities in the organization of written discourse in narrative and expository modes composed by native English, Chinese and Spanish college students. In addition, the cohesive devices were tallied to see whether there was distinction in terms of frequency and percentage of use across different language groups and modes. The researcher randomly selected participants registering in a remedial Basic Writing (BW) class or

31

in an English as a Second Language (ESL) class at college with equal number (n=30) from each language group regardless of the specific culture environment in which they were reared. The instruments employed to analyze the writing included: (a) the CUNY (City University of New York) Writing Assessment Test⁷ for the selection of participants; (b) the Milic logical categories ⁸ for the identification of the organizational patterns in writing and (c) Halliday and Hasan’s (1976) cohesion scale for determining the occurrence of cohesive elements in the written text. Through the calculation of ANOVA, the result revealed that different language groups displayed different organizational structures and their rhetoric patterns were consistent across modes. In terms of the total number of sentences each language group produced, more sentences were used in the narrative mode compared to the expository mode. Further, no significant difference was found in the percentages of the different Milic categories by three language groups when writing in two modes. With regard to the frequency of occurrence and types of cohesive devices, the referent category was not only the most frequently used by each language group across modes but was also more significant in each language in the narrative mode than in the expository mode after comparing the differences between the five categories of cohesive elements (e.g., reference, substitution, ellipsis, conjunction, lexical). Moreover, there were significant differences between the total number of cohesive devices used by each group but no significant differences in the percentages of devices employed by each group across modes. The findings of the study suggested that language background influenced one’s writing on organizational structure and the use of cohesive devices in the written language when writing in the narrative and expository modes. Norment’s

7 CUNY Writing Assessment Test (CWAT) is a writing assessment scale with six categories to place students in the Basic Writing or English as a second language program at the college.

8 The Milia Logical Categories was defined by Milic (1969) to identify sequential arrangement and relationship of sentences in a composition. It included eight categories: initial, additive, adversative, alternative, explanatory, illustrative, illative, and casual.

32

discovery in comparing different language group’s rhetoric was very enlightening;

however, the study is not without limitation. The participants he recruited, even having the same native language, were nurtured in different cultural environments of different countries. The specific cultural effect may affect one’s writing style or patterns. Finally, the comparison of cohesive devices in two modes was an initial study since the researcher only considered two factors—frequency and the total number—across languages and modes. More detailed analysis such as the comparison of subcategories of each cohesive device could be conducted.

Tierney and Mosenthal (1983) conducted a study in which twelfth grade students from two advanced composition classes were arbitrarily designated into two groups—the familiar and the unfamiliar group. All participants were required to write on two topics—a biographical sketch of Nathaniel Hawthorne and an essay on the theme of evil in Hawthorne’s work. The difference between the familiar and the unfamiliar group was that participants from the familiar group were shown the Hawthorne filmstrip while the unfamiliar group was shown the filmstrip of Henry David Thoreau prior to their writing. The content and structure of the two essays were held constant by the researchers’ outlines of the two topics and requiring students to follow the outlines. Next, participants’ essays were analyzed in terms of proportions of cohesive types and the ranking of the coherence in their writing by adopting Halliday and Hasan’s (1976) cohesive system. The result showed that text topics influenced the use of cohesion with regard to proportions of reference, conjunction, and lexical ties and that reference and lexical cohesion were the most powerful in discriminating cohesive pattern between topics. Higher proportion of personal and lexical ties in relation to total ties was found in the biographical text rather than the thematic texts. Further, a strong negative familiarity effect was found to correlate with relative coherence. The familiar group was mostly low-rated in the thematic essays.

33

As for the question whether or not a statistical account of cohesive ties could serve as an index of textual coherence, the researcher concluded that there seemed to be no causal relationship between proportional measures of cohesive ties within topics and coherence rankings within topics. To put it differently, a cohesion index was causally unrelated to a text’s coherence. (p. 228)

In 1987, Crowhurst conducted a study on the types of cohesion used in students’ writing by comparing differences between genres and students at different ages with different lines of development. In her study, students at grades six, ten and twelve were randomly assigned to write in one of the two modes—argumentation or narration. Through the procedure of randomization, thirty-five compositions were selected from each mode at each grade level for further analysis. The calculation of the number of cohesive ties differed from that of Witte and Faigley (1981). Instead of counting cohesive ties per T unit, Crowhurst computed the amount of cohesive ties per 100 hundred words and used a multivariate analysis of variance (MANOVA) to analyze the 12 dependent ties produced by previous calculation of cohesive ties.

Overall, no significant grade difference was found in the total use of cohesive devices.

Thus, the use of cohesive devices did not increase with age. However, with regard to lexical cohesion, significant differences were found in students at various grades. That is, there was an increasing use of synonyms and collocation with age. This result appeared to be consistent with previous studies done by Witte and Faigley (1981) and McCulley (1985) for it showed that greater use of synonyms and collocation correlated positively with better writing. In addition, there was a decrease in use of other types of cohesion with grade level such as exophora, causal conjunctive and temporal conjunctive. The increased or decreased use of lexical items or grammatical cohesive devices indicated mature writers had more diversified vocabulary and were capable of expanding and elaborating their thoughts. On the other hand, interactions

34

between the writing mode and use of cohesive ties were found. The cohesive ties employed varied differently with modes of writing in that students used pronominals, demonstratives and the, and temporal conjunctives more frequently in narrative prose than in argumentative one.

Similar research in relation to cohesion in writing was conducted by researchers Cox, Shanahan, and Sulzby (1990) as well. In their study, an attempt was made to investigate the relation between children’s reading performance and their use of cohesion in writing. Therefore, 48 L1students in the third- and fifth-grade in the elementary school from Chicago were randomly chosen to represent high and low reading comprehension groups respectively. Next, they were asked to write two narrative and two expository essays. The narrative task was about story telling;

therefore, before students started writing, three picture sets were showed and discussed in class. After students finished their first piece of narrative writing, they had to write a story using the same pictorial aids for other children at another school.

With regard to expository writing tasks, similar procedures were undertaken with the only difference being that articles were used to elicit ideas and discussion rather than picture sets. From the writing samples collected, the researchers first rated the appropriate or inappropriate use of cohesive devices and the overall cohesive harmony.

To address the question of whether grade interacted with reading level or not, ANOVAs was used. First, in terms of appropriate cohesive ties, older and more proficient readers demonstrated significantly more appropriate cohesive devices than poor and younger readers did. Thus, no significant effects were caused by the interactions of grade and reading level. Secondly, with regard to cohesive harmony, different results emerged in different genres. In narrative writing, both older and better readers had significantly higher scores in cohesive harmony than their younger and less skilled counterparts; while in expository writing, older students scored only

35

slightly more cohesive harmony than younger students, without any significant differences therein. No significant interactions were discovered. Thirdly, as for inappropriate cohesive ties, in narrative, the use of inappropriate cohesive ties significantly differentiated younger and poor readers from older and good readers; yet, interaction effects were not significant. In the case of expository writing, poor readers used inappropriate cohesive ties significantly more often than good readers but grade level caused no significant main effect. On the other hand, there was a significant interaction between grade and reading level. Finally, Pearson product moment correlation revealed that between frequency of appropriate cohesive ties and rank score for writing quality, in the case of narrative text non-significant difference was found; while significant difference was found for expository text. Additionally, significant difference was found between cohesive harmony and writing quality for both narrative and expository text. Consequently, in terms of the significant correlation between cohesive harmony and quality of writing, good readers tended to be good writers and achieve more cohesive harmony. The implication of the study was that cohesion knowledge was involved in both reading comprehension and writing quality regardless of genres or grade level and that cohesion knowledge lay in the process of developing reading skills and maturation.

In addition to the previous L1 research in writing, research in L2 EFL writing investigating cohesion in relation to genres can be found in Jin (2001) and Norment (2002). In Jin’s (2001) quantitative study of cohesion on Chinese graduate students’

writing, he analyzed six Chinese ESL graduate students in terms of the cohesion of their academic writing by adjusting Halliday and Hasan’s (1976) model to three main categories: coreference, conjunction, and lexical cohesion. The data collected was the participants’ writing assignments from advanced and immediate writing courses and these selected assignments covered three major academic genres: definition, analytic,

36

and research. In order to scrutinize students’ variation in the employment of cohesion in different genres, Jin first treated participants as a homogeneous group and applied the statistical tool, ANOVA, to see whether there were any differences in the use of cohesive ties. The result indicated that no significant difference was found in the use of subdivision of coreference, conjunction or lexical cohesion across genres. However, the mean of subcategories of coreference (e.g., personal, demonstrative, comparative reference) indicated that all three genres showed a strong preference for personal reference and demonstrative reference and that the means of personal reference and demonstrative reference in the analytic and the research papers were higher than in the definition paper. As for the use of conjunctions, the definition and analytic paper had a close mean while the research paper had a much or slightly higher mean in the category of additive and temporal conjunction respectively. With regard to lexical cohesion, four subcategories of lexical cohesion were examined: same word, synonym, super-ordinate, and general word. Similar to the results obtained from the grammatical cohesive ties, none of their frequencies revealed significant differences in three genres.

However, a fact worth our attention was the significantly higher mean of same word compared to the three subcategories, which suggested that Chinese students have a strong preference for repetition of the same word as a lexical cohesive device.

After comparing the use of cohesive devices in the three genres, Jin explored the issue of whether students of different proficiency groups would perform differently in the subsection of grammatical and lexical cohesion. In this way, Jin eliminated the differences of genres by viewing them as a homogeneous genre and separating students based on the writing course — advanced and intermediate. After the application of T-tests, significant differences were found to exist in the use of personal reference, temporal conjunctions, and synonyms. These differences in cohesion showed that advanced writers endeavored to familiarize the readers through

37

the use of personal reference and that advanced writers preferred sequential organization of ideas and demonstrated higher sophistication in writing than intermediate student writers. Additionally, from the observation of means, advanced writers overall used more conjunctions than intermediate writers and both proficiency groups favored using lexical repetition to achieve textual cohesion. Finally, a two-factor split plot ANOVA was operated to test the overall variation of cohesion across genres and proficiency levels.

The result showed that significant differences were found in overall cohesion between the two proficiency groups and there were significant differences among the three genres, but no significant differences in the interaction between proficiency and genre. From the results of the study, Jin also suggested that overall language proficiency and writing ability are somehow correlated because different proficiency resulted in differences in students’ cohesiveness in their writing. Even though Jin’s finding was quite informative, the design of the study is not without flaws. In the comparison of cohesion in different genres, Jin treated all students as a homogeneous group and analyzed their writing, which was quite unreasonable in terms of students’

different proficiency levels and years of stay in the United States. Second, the classification of students into different proficiency groups by the placement of advanced or intermediate course was questionable because scores of TOFEL tests showed that one intermediate student scored as high as one advanced student. Finally, we should also be reserved about any conclusion based on the limited number of participants in the present study.

Norment (2002) conducted a quantitative and qualitative analysis of textual cohesion in African American students’ writing in narrative, argumentative, and expository modes. In his study, he randomly selected 30 African American college-aged students enrolling in a remedial basic writing (BW) or a freshman

38

composition (FC) writing class from City University of New York. The participants were required to produce three different modes of writing—narrative, argumentative, and expository writing—over a span of six weeks. The collected writing samples were rated and further divided into high and low groups with fifteen essays from each group. Next, cohesive elements in student essays were analyzed adopting Halliday and Hasan’s (1976) taxonomy of cohesive system by six raters and inter-rater reliability reached a median of 0.91, revealing great consistency across raters.

Statistical Package for the Social Sciences (SPSS) computer program was applied to carry out all the calculations. With modes and proficiency levels as independent variables, 3 X 2 ANOVA was utilized to see whether there was any interaction between these two factors.

The results showed significant differences between the total number of cohesive devices used within the group but no significant differences in the percentages of devices used across modes. As for cross-modal comparison of cohesive devices in students’ writing, in each mode lexical devices were the most frequently used while ellipsis devices were the least frequently used. Further, lexical and conjunctive cohesive devices were used more frequently in narration than in argumentation and exposition. Similar frequencies of usage in conjunctions and references occurred in exposition and narration. A significantly higher frequency of occurrence of endophora in narration and argumentation was found as well. On the other hand, high-rated writers distinguished from low-rated writers by using a substantially higher percentage of cohesive devices, which suggested that high proficiency writers tended to establish cohesive bonds in text more than their low proficiency counterparts did.

Additionally, high-rated writers also had more words and denser ties per sentence than low-rated writers. The findings of the study revealed that different modes of writing and writers of different proficiency levels may influence the pattern and frequency of

39

cohesive devices in written discourse. Norment’s detailed analysis of African American writers’ performance in different genres was informative. However, his presentation of the statistical analysis was not without drawbacks, for the report does not quite match the data presented in the tables, thus rendering the results less reliable and valid. In addition, the researcher did not explain what the difference was between repetition and the same item within his lexical categories of cohesion. More importantly, and inconsistently, since quantitative analysis was present, analysis of the qualitative was invisible in the study.