ภัทร์ธีรา เทียนเพิ่มพูล. การประยุกต์ใช้การวิเคราะห์แบบอิงเนื้อหากับงานเขียน. Master's Degree(Applied Linguistics for English Language Teaching). มหาวิทยาลัยเทคโนโลยีพระจอมเกล้าธนบุรี. : King Mongkut's University of Technology Thonburi, 2546.
Abstract:
Since writing is a productive skill and it is freely expressed according to the writers schemata, it is fairly difficult to evaluate some aspects of writing. One aspect of writing which is problematic is organisation. There are two areas of organisation: cohesion and coherence. Cohesion is explicitly shown in the texts. On the other hand, coherence is not shown on the surface of the text so most teachers seem to pay attention to cohesion rather than coherence in marking organisation. Moreover, there are no objective criteria for marking coherence. Instead, teachers usually use their own intuition in marking coherence. The difficulties in marking coherence may lead to both an imbalance in marking organisation and a lack of reliability in marking. Because of this, this study is carried out to investigate a method of analysing coherence, which is topic-based analysis. Topic-based analysis is chosen as the key method of analysis because it is claimed that it is fairly reliable and objectively measures coherence. Therefore, topic-based analysis has been applied to writing in order to see its applicability and reliability. To apply topic-based analysis to writing, the procedures consist of three main stages: preparing data, identifying key concepts, and identifying hierarchical relationships. To prepare the data, ellipsis, referents, T-units, and repetition are identified. After that, nouns or noun phrases which occur twice in separate T-units are identified as a key concept. Next, the relationships between these concepts are identified using the logical relationship of implication, namely, hyponymy, meronymy, entity-characteristic, possessor-possessed, and cause-effect. Line diagrams representative of schemata can then be drawn up and topic development through the discourse can be presented using these diagrams. After following the methods of analysis, the results are shown in terms of distances of moves which is based on numbers of logical links between two concepts. A distance of moves which is higher than two shows a discontinuity in the discourse or a coherence break. After this, the coherence breaks identified by topic-based analysis are presented according to distances of moves and the results of topic-based analysis are interpreted as scores. There are four ways of calculating scores: mean distance of moves, standard deviation, percentage of coherence breaks, and number of moves per ten T-units. There are some connections between these scores because the results of the four methods of scoring are fairly similar. Having scores interpreted by topic-based analysis relating to each other does not enable us to state that topic-based analysis is reliable. To investigate the reliability of topic-based analysis in writing, the scores were compared with the teachers marking for coherence and topic-based analysis identification of coherence breaks was also compared with that of teachers. From the comparison, there are some agreements between topic-based analysis and the teachers, but the level of agreement are not very clear. There are three factors affecting these dissimilarities: unreliability of teachers marking, teachers perspectives in identifying coherence breaks, and nature of language. After considering these factors, the reliability and validity of topic-based analysis are investigated. The results show that topic-based analysis is as reliable as a teacher. Furthermore, it seems likely that topic-based analysis is actually measuring coherence.