學生答題寫作特性及其與表現之相關研究
No Thumbnail Available
Date
2007
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
本研究以系統功能語言學為理論基礎，分析我國學生在TIMSS 2003生命科學的推理分析自由作答題中，語法單位、語法結構、語篇功能與推理情形三大類共十九項答題特性的表現，並探討答題特性間的相關、答對學生和答錯學生在科學語言使用上的答題特性差異，進一步由答題特性建立成績對錯的預測迴歸模型。研究對象為我國參與TIMSS 2003實測的八年級學生，研究資料扣除未作答資料後，依答對學生和答錯學生的資料比例，分層隨機抽樣自TIMSS 2003的原始學生回答內容，共取得1037筆進行統計分析，研究發現如下：
（1）答對學生、答錯學生與整體學生的十九項答題特性現況。
（2）答對學生的作答特性顯著優於答錯學生，僅在科學詞彙密度、零代詞總數、扣除無推論情形之作答後的小句內推論數三項未達顯著差異。
（3）答對學生和答錯學生作答特性間的差異，反映了進行科學問題的文字作答時，答對者使用的是特殊的科學語言使用方式。
（4）由學生答題特性對成績對錯進行的預測準確率為87.8％，迴歸預測模型為：
Logit（y）＝－4.351－0.058×字數＋1.198×小句數
＋0.753×平均名詞組深度＋0.987×平均小句深度
－0.504×零代詞總數－2.370×不可還原零代詞數
＋1.997×正確推論事件數－2.757×錯誤推論事件數。
Drawn on the perspective of Systemic Functional Linguistics (SFL), this study aimed at analyzing students' writing characteristics of answering TIMSS 2003 constructed-response items, the relations among students' writing characteristics, and the differences of students’ writing characteristics between correct and incorrect responses, and further created a regression model on foregoing findings to predict if the response of an item is correct. The quality and quantity of students' writing responses were analyzed in terms of the syntax units, the syntax structure under the ideational function of SFL, and the textual function of SFL, nineteen writing characteristics in three categories in all were investigated. After getting rid of blank response data, one thousand and thirty-seven data from the responses of Taiwan 8th grade students participating in TIMSS 2003 main survey were stratification randomized by the percentage of correct and incorrect response data and analyzed statistically. Major findings were as followed : (a)The current conditions of nineteen writing characteristics of TIMSS 2003 entire, correct, and incorrect students’ responses. (b)The writing characteristics of correct students' responses were better than incorrect ones, but there was no significant difference between correct and incorrect students' responses in the three aspects of lexical density, the number of zero pronouns, and within-clause reasoning getting rid of responses with no reasoning. (c)The difference of writing characteristics between correct and incorrect response reflected that students whose answers were correct used a specific way of science language. (d)The prediction accuracy of the model created by students' writing characteristics was 87.8%, and the regression model was as followed : Logit（y）＝－4.351－0.058×[the number of words] ＋1.198×[the number of clauses] ＋0.753×[the average depth of nominal groups] ＋0.987×[the average depth of clauses] －0.504×[the number of zero pronouns] －2.370×[the number of irretrievable zero pronouns] ＋1.997×[the number of correct reasoning] －2.757×[the number of incorrect reasoning]
Drawn on the perspective of Systemic Functional Linguistics (SFL), this study aimed at analyzing students' writing characteristics of answering TIMSS 2003 constructed-response items, the relations among students' writing characteristics, and the differences of students’ writing characteristics between correct and incorrect responses, and further created a regression model on foregoing findings to predict if the response of an item is correct. The quality and quantity of students' writing responses were analyzed in terms of the syntax units, the syntax structure under the ideational function of SFL, and the textual function of SFL, nineteen writing characteristics in three categories in all were investigated. After getting rid of blank response data, one thousand and thirty-seven data from the responses of Taiwan 8th grade students participating in TIMSS 2003 main survey were stratification randomized by the percentage of correct and incorrect response data and analyzed statistically. Major findings were as followed : (a)The current conditions of nineteen writing characteristics of TIMSS 2003 entire, correct, and incorrect students’ responses. (b)The writing characteristics of correct students' responses were better than incorrect ones, but there was no significant difference between correct and incorrect students' responses in the three aspects of lexical density, the number of zero pronouns, and within-clause reasoning getting rid of responses with no reasoning. (c)The difference of writing characteristics between correct and incorrect response reflected that students whose answers were correct used a specific way of science language. (d)The prediction accuracy of the model created by students' writing characteristics was 87.8%, and the regression model was as followed : Logit（y）＝－4.351－0.058×[the number of words] ＋1.198×[the number of clauses] ＋0.753×[the average depth of nominal groups] ＋0.987×[the average depth of clauses] －0.504×[the number of zero pronouns] －2.370×[the number of irretrievable zero pronouns] ＋1.997×[the number of correct reasoning] －2.757×[the number of incorrect reasoning]
Description
Keywords
系統功能語言, TIMSS, 認知領域, 自由作答題, 答題寫作特性