Abstract：Based on Rhetorical Structure Theory （RST）， this paper aims to investigate into the typical features of rhetorical relations and hierarchical structure in College English Argumentative writing and their relationship with writing quality with a case study. The research findings shows that theres no major difference in terms of total numbers and types of rhetorical relations between groups of high score and low score. However， there are differences among the most frequently used relations by the two groups and group of low score has more unclear rhetorical relations than group of high score.

Key words：Rhetorical Structure Theory； rhetorical relations； argumentative writing

ⅠIntroduction

Rhetorical Structure theory （RST） was proposed by Mann and Thompson as an explanation of coherence relations of texts and now has been widely used in discourse related studies. RST is the theory with which we can explain the structural relations of a text， identifying the relations by which one span of text can be related to another rhetorically， thus developing the text. As we all known， text coherence is one of the critical standard for reviewing whether a writing is good or not， therefore the linguistic approach to realize text coherence in student writing should be given special attention for language teachers. This paper intends to investigate into the typical features of rhetorical relations and hierarchical structure in College English argumentative writing in order to obtain the current situation of structural conception in students writing. Because of the limitation of space and time， we will take students of Northeast Normal University for the case study. In the study， 4 college student argumentative writings （2 with the highest scores in the class and 2 with the lowest scores in the same class） which fall into two groups （Group H and Group L） were chose as the samples. We will then further explore the relationship of rhetorical relations with writing quality in order to testify the contribution of RST in construing writing.

Ⅱ Literature Review and Theoretical Framework

Rhetorical structure relations are central constructs in Rhetorical Structure Theory （RST）， which was developed by Mann and Thompson in a paper presented in 1978 （Lin & Yang， 2007）. Since that publication， the theory is used in Linguistics as an explanation of coherence relations of texts and has become one of the most widely applied discourse theories （Marcu， 1997）. RST is an approach to text coherence which can be used to describe rhetorical relations and explore the relational structure of any relative coherent text by firstly dividing the text into some minimal units of interest. For Mann and Thompson， the size of the “minimal units of interest” of text analysis is arbitrary. As long as they have independent functional integrity， the minimal units can be clauses or larger units composed of clauses. A role in the text is then assigned to each such unit， primarily by linking parts of the text together using relations， and by aggregating related parts into spans， so that the text is connected together into a hierarchic structure （Lin & Yang， 2007）. As Mann and Thompson （1992） proposed， a text span is any portion of text that has an RST structure （and thus has a functional integrity， from a text-organizational point of view）， or that is realized by a unit. Two spans of text enter into a rhetorical relation such as elaboration， cause， circumstance or motivation. The relation is typically asymmetric； one span is a nuclear and the other span is a satellite. According to Mann and Thompson， the nuclear-satellite relation is the major structure in a text which is identified meaning-based rather than structurally. Nevertheless， RST also identifies multinuclear relations such as contrast. Rhetorical relations can be applied at all levels of discourse. It is hierarchical and the analysis follows a top-down manner.

Mann and Thompson （1978） presented a list of 23 rhetorical relations. The Joint relation later appeared in Mann and Thompsons paper is also necessary. So all in all， a list of 24 rhetorical relations is the RST we applied in paper.

Descriptive RST lays a foundation for studies in contrastive rhetoric. Cui's analysis of Mandarin and English essays （Cui， 1985） is an example. RST has also proven to be useful in analyzing narrative discourse as well. Kumpf （1986） is a study of the interlanguage of Japanese and Spanish speakers. The author shows that RST is valuable in describing the grammatical and rhetorical properties of the narratives produced by these speakers. Since the coherence of a text depends in part on the Relational Propositions， RST has been useful in the study of text coherence. Though it has been studied a lot， few researches focus on the Chinese EFL texts written by Chinese speakers （Mann and Thompson， 1992）. The purpose of the study is that， on the one hand we will see the feature of EFL learners writing in terms of rhetorical relation realizations， and on the other hand， try to produce some pedagogical implication in teaching the texture organization of writing.

Ⅲ Research Methodology

This is a case study of an university In northeast of China. With stratified-cluster random sampling， 4 argumentative writings （2 with the highest scores in the class and 2 with the lowest scores in the same class） were chosen as the samples， which fall into two groups： Group H and Group L. Group H contains two samples， Sample A and Sample B. Group L contains Sample C and Sample D. These 4 texts were written by first-year college students to address a same writing topic Publicizing Lists of Uncivilized Residents. These writings were rated automatically by computers on a TRP system and they were done by students after class. The word count of the sample texts are all about 150 words， ranging from 145 to 157 words. Each text is cut into minimal discourse units for rhetorical relation analysis. For convenience and clarity and in accordance to previous studies of RST， the minimal units， or text span in the analysis is “idea group sentence” which is numbered in each text. “Idea group sentence” means the unit can be a clause or a sentence or more than a sentence which mainly depending on whether the meaning the unit convey belongs to an unity.

Ⅳ Data Analysis and Discussion

When the two groups and four samples are compared， three points are quite clear. The first point seems to be very evident while researcher analyzing the sample texts and drawing Rhetorical relations and Hierarchical structure figures， that is， it seems that Group L has more unclear rhetorical relations which have been annotated by dotted lines than Group H. It means the logic and structure in Group L is less apparent and the writings are less consistent. For example， in the figure 4 there are two dotted arrows presenting unclear or uncertain rhetorical relations and in figure 5 there are one dotted arrow. Secondly， one phenomenon need to be noted is that writings in Group L have less hierarchies. Sample C has only 4 hierarchical layers while both two samples in Group H own 5 hierarchical layers. However， as the data is far from enough to be sure to give the conclusion， this point need further more investigation. We have summarized a table to show the contrast between Group H and Group L in terms of types and times of rhetorical relations and thereby analyzing the underlying features and testify if there is a relationship between quality of writing and its rhetorical relations in texts. The results are shown in table 1.

From the table， we can summarize the third point of the findings. In both two groups， contrast relation is of the highest appearance rate. This shows that probably all argumentative writings no matter with high score or low score like using contrast relations to form their argument. This may be interpreted referring to genre as all four sample are all argumentative writings. Besides， after comparison， it can be found that writings in Group H possess more types of rhetorical relations while writings in Group L tend to have a high frequency use of certain rhetorical relations such as contrast， interpretation and joint relation.

ⅤConclusion

The research findings shows that theres no major difference in terms of total numbers and types of rhetorical relations between groups of high score and low score. But some rhetorical relations， like contrast relation， have much higher frequency in both groups. The reason may be the specific feature of argumentative writing and specific directions or topic of the writing. Whats more， there are differences among the most frequently used relations by the two group and group of low score has more unclear rhetorical relations than group of high score.

Reference

[1]Cui， Songren. Comparing Structures of Essays in Chinese and English[D]. UCLA， 1985.

[2]Kumpf， Lorraine. Structuring Narratives in a Second Language： a Description of Rhetoric and Grammar[D]. Los Angeles： University of California， 1986.

[3]Mann W C， Thompson S A. Rhetorical Structure Theory： Description and Construction of Text Structures[M]// Natural Language Generation. Springer Netherlands， 1987：85-95.

[4]Mann W C， Thompson S A. Rhetorical Structure Theory： Toward a functional theory of text organization[J]. Text & Talk， 1988， 8（3）：243-281.

[5]Mann W C， Thompson S A. Discourse description ： diverse linguistic analyses of a fund-raising text[J]. Language， 1992， 69（3）.

[6]Mann W C， Matthiessen C M I M， Thompson S A. Rhetorical structure theory and text analysis[J]. Nasa Sti/recon Technical Report N， 1992， 90.

[7]Thompson G. Introducing functional grammar[M]. 外語教学与研究出版社， 2012.

[8]林伟，杨玉晨. 英语与篇分析[M]. 上海：复旦大学出版社，2007.

作者简介：王安琪，女，汉，1993.07，黑龙江大庆人，东北师范大学，硕士在读，研究方向：语篇分析。