以測驗和眼動型態建構交互作用效果的圖文閱讀理解模式

Abstract

本研究以眼動資料與測驗結果，建構交互作用效果的統計圖文閱讀理解模式。論文中包含三個研究。研究一發現交互作用效果對於一般成人而言是複雜的概念，不同領域專家的看法也不完全一致。但交叉型態的圖示能使讀者連結到「交互」的字面意義，協助判讀。因此，研究二以交叉統計圖的有無為實驗操弄，發現統計圖確實影響閱讀理解的歷程和效果，但提升幅度未如預期的佳。於是，研究三以圖文訊息對應性和圖形對稱性的觀點，修改統計圖，並以讀者閱讀理解表現與眼動資料討論圖文整合的歷程。 研究一包含兩個部分，1-1探討一般民眾對交互作用效果的解讀，研究對象為12位大專（含）以上，包含研究所與社會人士之民眾，不限領域。研究先提供受訪者一個交互作用效果的範例，接著請他們判斷後續兩個文本的現象是否為交互作用效果，並解釋原因。結果發現，讀者的解讀多侷限在交互作用的字面意義，且認為單純主要效果的描述相當難理解，而略有基礎統計概念者可簡略描述資料分析方式；此外，讀者會將交叉圖形連結交互一詞進行解釋。研究1-2藉由領域專家獲得交互作用效果的重要概念內涵，以編製後續實驗材料，研究對象為醫藥、物理、心理計量三個領域，共16位領域專家。研究材料是兩個交互作用效果文本，受訪者需判斷該文本是否為交互作用效果，並以口頭與繪圖說明之。結果顯示，心理計量專家著重於變項概念與統計考驗的重要性，並運用統計學交互作用效果的概念，進行跨文本的解釋，其他領域專家則是針對專業領域中交互作用的運作機制進行論述。最後，整合研究一的結果，得出交互作用效果包含基本的和進階的概念。 研究二的研究對象為60位大學生（含）以上的社會人士，探討純文與圖文並置對讀者交互作用效果的閱讀理解之影響。於正式眼動實驗前，以301位大學生（含）以上的社會人士進行團測，以確立實驗材料的難度是否適合一般大眾閱讀，並掌握讀者所需的閱讀時間以利設計正式實驗。研究二採單因子受試者間設計，實驗組閱讀圖文並置的版本，控制組則閱讀純文字的版本。閱讀時以眼動儀記錄眼動軌跡，兩組在閱讀完實驗材料後，都接著進行閱讀理解測驗。結果發現，多了統計圖後，提升了讀者整體的閱讀理解正確率，特別是基本概念的題目，亦即藉由統計圖座標軸標示的變項可以協助讀者讀取表層特徵。眼動追蹤資料顯示，多了圖之後，讀者的閱讀時間並未增加，而是分配時間去讀圖，節省讀文的時間；但圖文版的讀者在句末較少進行語意統整，可能因抱持仰賴圖的態度，而未投入與純文組同等程度的努力去讀文。 研究三的研究對象為70位大學生（含）以上的一般成人。實驗設計採單因子受試者間設計，實驗材料是沿用研究二的交互作用效果的文字內容，以圖文訊息對應性與圖形對稱性為實驗操弄，修改統計圖的編製方式，共有三種實驗材料版本—非對稱、對稱和單圖。非對稱組和對稱組的文本中都有兩張圖，差別在於兩張圖的交叉開口是否對稱；單圖組有一張交叉開口對稱的圖（同對稱組的其中一張圖）。結果發現，（一）對稱組與單圖組在閱讀理解正確率無顯著差異，但兩組的眼動型態有所差異，顯示圖文訊息對應性影響閱讀歷程但對閱讀理解的影響不大。相較於單圖組在讀完文之後始去讀圖，兩張圖會促使讀者在閱讀歷程中途（段落2和段落4）遇到圖文對應的訊息就去讀圖。推測其因，雙圖提供較完整的訊息且也較為醒目，吸引讀者在尚未讀完文時就去讀圖。（二）非對稱與對稱在閱讀理解正確率無顯著差異，但兩組的眼動型態亦有所差異，顯示圖形對稱性影響閱讀歷程但對理解的影響不大。非對稱組花較多的時間讀文，且有更多跨圖間的閱讀。即圖形兩邊開口不對稱易引發讀者思考該表徵的意義，而增加讀文的時間，且在圖間來回搜尋以比對兩張圖的差異。此外，閱讀過程中，非對稱組會同時參照二圖，反之，對稱組僅到所讀到的段落之鄰近圖示閱讀。即非對稱組視兩張圖為獨立的，需同時參照並比對雙圖的差異來確定圖文訊息是否對應；然而，對稱組則視雙圖為相同的圖形樣式，因此採取便利的閱讀路徑，僅到靠近文的圖上搜尋圖文對應的訊息。（三）非對稱與單圖的比較旨在分析兩種實驗操弄加乘的效果，結果非對稱組在基本和進階概念的理解都有較佳的傾向。 本研究以交互作用效果統計圖文的版置概括有無圖示、圖文訊息對應（雙圖與單圖），以及圖像知覺（非對稱與對稱）等實驗操弄，提出統計圖文閱讀理解的模式。相對於純文的閱讀，加入統計圖會使讀者從圖文參照中，讀取圖所蘊涵的語意和空間表徵，形成他們對交互作用效果的基本概念，也節省讀文的時間。其次，雙圖的知覺醒目性會引導讀者在閱讀文章中途就去參照圖，再繼續處理文。最後，統計圖形的不對稱性，會影響讀者的閱讀時間與圖文整合的型態。

關鍵字：交互作用效果、統計圖文、閱讀理解、眼動型態、圖文整合This study composing of three researches aimed to construct a model of reading comprehension of interaction effect with statistical text and figure. First, it was found in the Research 1 that interaction effect is a complicated concept for the public and field experts hold diverse perspectives of it even though. Yet, the cross-over figure could prime readers to associate the shape of cross-over with the surface meaning of “interaction”. Hence, adding a cross-over figure of interaction effect was used as the manipulation in the Research 2, which was explored to influence the process and the performance of reading comprehension. Nevertheless, the effect of adding a cross-over figure was not as great as the expected. Accordingly, consistency of information between figure and text and symmetry of figure were operated in the Research 3 to explore the effect on participants’ performance and process of reading comprehension. Two sub researches were involved in the Research 1. Research 1-1 aimed to explore how the public comprehended interaction effect with 12 participants. Participants would read an interaction effect example and then they were asked to interpret interaction effect in the following twotests and make explanations. It was revealed that participants’ interpretation was the surface meaning of interaction and they claimed the difficulty of description of simple main effect. Furthermore, some participants with basic statistical background knowledge could make simple explanations of data analysis. Research 1-2 aimed to come by the important concepts of interaction effect to make the following experimental materials. 16 participants were field experts of medicine, physics and psycho-statistics. Participants would read two text of interaction effect and then interpret interaction effect both with verbal explanations as well as drawing figures. It was found that experts of psycho-statistics focused on the concept of variable and test significance and were inclined to utilize general principles of statistics to interpret different text, meanwhile experts of other fields emphasized on the mechanism of interaction effect. Last, fundamental and advanced concepts of interaction effect were obtained. The Research 2 aimed to explore the effect on participants’ reading comprehension of interaction effect from text and text with figure. There were 60 participants. A group test was held with 301 participants to confirm the readability and the difficulty of experimental materials and to design the experimental procedure before the formal experiments. One-factor between groups was used in the Research 2 with text group as control group and text with figure group as experimental group. EyeLink 2000 was utilized to record participants’ eye movement during experiments. It was explored that adding a statistical figure promoted participants’ accuracy of performance of reading comprehension especially on items belonging to fundamental concepts, which revealed the effect of statistical figure on benefitting participants to grab the surface representations from figure. Total fixation duration were not on increase; participants distributed reading time on figure to save the reading time on text. For another thing, text with figure group performed less semantic integration in the end of sentences, which could refer to it that participants in this group consumed less mental effort on reading text for they had intention to depend on figure. The Research 2 aimed to realize how the effect of manipulations, involving consistency of information between figure and text and symmetry of figure impacted on participants’ reading comprehension. There were 70 participants. Experimental materials were text with two unsymmetrical cross-over figures (experimental group 1), text with two symmetrical cross-over figures (experimental group 2), and text with one symmetrical cross-over figure (baseline group). Results are shown as follows. First, there is no test significance of performance of reading comprehension between experimental group 2 and baseline group. However, eye movement between these two groups revealed differences on the process of reading comprehension that experimental group 1 would read the figures during reading text meanwhile experimental group 2 would read the figure after reading the whole text. Second, differences of eye movement between experimental group 1 and baseline group showed that unsymmetrical figures facilitated participants to contemplate the meaning of this representation so as to spend more time on text and eye movement across figures. Additionally, experimental group 1 would refer to two figures simultaneously for they regarded two figures independent. Yet, experimental group 1 would trace back to the figure just next to the read text for they held two figures independent were the same. Last, the effect of two manipulations would influence enhance both basic and advanced concepts of interaction effect. Composition was concerned as the integrative concept to combine all the manipulations in this study so as to construct the model of reading comprehension of statistical text with figure. Semantic and spatial representations of figure benefited readers to form concepts of interaction effect and save the reading time on text. Salience of figure guided readers to refer to figure during reading text. Finally, symmetrical and unsymmetrical figures would influence the reading time and integration of figure and text. Keyword：interaction effect, statistical figure and text, reading comprehension, eye movement, integration of figure and text