兒童局部與整體連貫性推論能力發展研究

A developmental study of children's local and global coherence inference abilities

理解並非一個單一的歷程，相反的，理解需要許多種推論能力以建構情境表徵。推論的發展差異在閱讀研究中是重要議題，然而現階段對於局部與整體連貫性推論的發展順序尚待釐清。局部連貫性推論需要建立字詞與句之間的連結（例如代名詞），整體連貫性推論需要讀者超過文本並結合自身知識已填補文章中缺失的細節。本研究針對局部與整體連貫性推論進行年級間的差異分析，也進一步以群聚分析探討跨年級資料合併後的潛在型態。研究採用六篇各約200字的短篇故事，每篇有1題字面題，2題局部連貫性推論題及1題整體連貫性推論問題，皆為選擇題形式。結果發現，四、五年級學童在兩種推論類型表現都優於三年級學童，且五年級學童在局部連貫性推論的表現優於三、四年級學童；然而此差距並不必然是年紀的影響，跨年級的學童也可能存在有共同的反應型態：即使年齡、能力相當的學童中，仍在局部與整體連貫性的推論有不同的表現。群聚分析得出四種不同的反應型態，顯示理解所需的推論技能並非都同步發展。對能夠做整體連貫性推論的學童來說，並非皆具有高局部連貫性推論能力。本研究結果顯示出學童在推論發展順序上存在個別差異，應針對學童所欠缺的推論能力進行教學，給予更合適的推論課程。

Comprehension is not a unitary process. Instead, in addition to many other cognitive skills, several types of inference abilities are required to construct a situation model from the text. Despite knowing developmental differences in students' inference abilities for reading comprehension, definite conclusions on the relative developmental sequence of local and global coherence inference cannot be reached. For reading comprehension, local coherence inferences are needed to establish links between words and sentences (e.g., anaphor). Global coherence inferences are required for readers to go beyond the text and integrate it with their general knowledge to fill the missing details and understand the text as a whole. In this study, we examined the development of both local and global inferences between and across grade levels. In this study, six short narrative passages were selected, each constituting approximately 200 Chinese characters. We devised one literal, two local, and one global coherence inference questions for each text in the multiple-choice format. The results clearly showed that both fourth and fifth grade students performed better than third grade students for both types of inferences. Additionally, fifth grade students had higher scores than third and fourth grade students for local and global coherence inferences. Cluster analysis of the inferences across grades indicated that students' performance was not solely determined by their grade. Individual differences within grades and similar patterns across grades were observed. Cluster analysis results showed the existence of four groups. The results confirmed that comprehension is a complex process and that not all subskills can be developed simultaneously. Not all students with high global inference ability are good at making local coherence inferences. Individual differences in the patterns of inference abilities are revealing and call for more research attention. The patterns revealed in this study can also serve as good foundations for reading instructions that emphasize individual differences.