應用討論式學習於程式設計學習─一個初步的實驗與探討

很多人認為將傳統教學的實體教材轉變成可用電腦觀看與傾聽的文字、圖片、影音等多媒體，在網際網路上傳播，運用科技造就學習的便利性，就是所謂的學習科技，尤其強調線上學習與遠距教學。但我們認為學習科技的重點在於研究與發展幫助人們學習的科技，經由使用這種幫助，學習者的學習成效就能有明顯地提升。任何科技的研發，只要符合這個條件，就是學習科技。我們可以歸納出獲得解決新問題的方法是從對問題形成瞭解、應用已有的知識來解決問題、與不同的意見相互交流、反省自己的認知、修正對問題的瞭解，再繼續應用重新獲得的知識來解決問題，如此不斷的循環直到解決問題。我們所要探討的是如何適當的使用電腦技術與網路技術，在課堂上幫助學生作即時的、有控制的集體討論，討論的方式將是(1)具強迫性，(2)可個別assign題目，(3)學習者之間彼此匿名、彼此評估，(4)學習者可觀看彼此的答案，(5)老師與助教可察看學習者的作答記錄、作歷程評量。為幫助學習者學習程式設計，我們發展了一套討論式程式設計學習系統（Discussion on Programming Learning, DPL），這套系統是給予學習者相同的題目，根據程式題目對學習者提出問題，要求所有學習者參與問題的討論，並對其他學習者的回答作批評與回應，使學習者能真正瞭解程式是如何運作的，在每天課程的最後，學習者還要進行程式作業的寫作。本實驗以中原大學九十三學年度大學二年級的學生為實驗對象，總計人數有111人，實驗組11人，大對照組96人，小對照組4人。實驗組接受DPL的訓練，大對照組與小對照組則無接受DPL的訓練。實驗過後的結果與分析如下： 1. 實驗組經過討論式學習的訓練過後，後測成績(相較於前測成績)有顯著的進步。 2. 小對照組後測成績(相較於前測成績)沒有顯著的進步。 3. 大對照組後測成績(相較於前測成績)沒有顯著的進步。實驗組的平均後測成績仍不及大對照組，而且經ANOVA信心度95%的檢定，二者有顯著差異。 4. 實驗組的平均成績勝過小對照組，但經ANOVA信心度95%的檢定，實驗組的後測成績與小對照組的後測成績並沒有顯著差異。 5. 由1.與2.來看，本實驗有成功的可能，但因樣本數不足無法做明確的結論。 6. 實驗組平均後測成績仍不及大對照組，其原因之一應該是實驗組的平均專心度(0.644)未達認真程度應有之參考標準(0.82)，事實上實驗組認真程度最高者也只有0.75。

關鍵字

學習科技；程式設計；討論式學習

並列摘要

With the advance of computer and networking technology, it is now possible to create learning technology that can make significant changes to how human learn. What we propose in this thesis is a methodology for remedial learning. The methodology, called learning by discussion, has its root in constructivism. The basic idea of learning by discussion is that learners first present their ideas of the topic. Then, by evaluating answers presented by other people, each learner has a chance to revise his/her own ideas. This way, learners may form a consensus. The formation of such a consensus can be viewed as something that is constructed by the learners themselves. In this thesis, we explore the use of a computer technology that facilitates learning by discussion, and we focus on remedial learning of introductory programming. The proposed learning methodology can be characterized in four ways. One, it is forced discussion. Two, different problems may be assigned to different individuals when needed. Three, learners evaluate one another's work in an anonymous way. Four, the instructor (or the assistants) can evaluate the learning process of each learner. An educational experiment was performed during the fall semester of 2004. Eleven students participated in the experiment. Their performance was contrasted with a small control group of four. The reason that there were so few subjects involved was that it was remedial learning for those who fail the normal introductory programming course. The experiment led to two main findings. The experimental group outperformed the control group in making progress (in exam scores). However, the experimental group did not make enough progresses as we would like them to, and our analysis of the collected data suggested that the subjects probably did not put in enough efforts in their learning processes.